Skip to content

Instantly share code, notes, and snippets.

@worthyag

worthyag/Can-machines-understand-how-you-feel.ipynb

Created February 15, 2024 07:31
Show Gist options
  • Save worthyag/215693a2bb403c0cf1a121296911302b to your computer and use it in GitHub Desktop.
Save worthyag/215693a2bb403c0cf1a121296911302b to your computer and use it in GitHub Desktop.
Download ZIP
Can machines understand how you feel?
Display the source blob
Display the rendered blob
Raw
Can-machines-understand-how-you-feel.ipynb
{
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "N8DVaLzZTKHk"
},
"source": [
"(CM3015) Machine Learning and Neural Networks - Following the universal workflow of DLWP 4.5 (1st Edition)"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "P6tbmyP8TyH5"
},
"source": [
"# Can machines understand how you feel? Using machine learning models to predict sentiments."
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "a6hutTq8VjWb"
},
"source": [
"*Aiming to predict the sentiments of unseen data.*"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "8xhXc8-kH0Ti",
"tags": []
},
"source": [
"# 1 Defining the problem and assembling a dataset"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "GeQEdDNPWXCk",
"tags": []
},
"source": [
"## 1.1 Introduction and background"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Sentiment analysis, defined as being \"*the process of analyzing digital text to determine if the emotional tone of the message is positive, negative, or neutral*\"[1], has been a subject of research for many decades. However in recent years, there have been many breakthroughs and significant advancements. What triggered this? Why are people becoming increasingly interested in this particular domain?\n",
"\n",
"The reason would be the sheer use cases it provides. For instance, it allows for business insights, as it enables companies to gain insights into customer opinions, preferences, and the like [2]. This information can then be used to improve products and develop marketing strategies. It is also used in the political field, allowing analysts to gauge public opinion on political candidates and issues [3]. In addition, it is used in the medical field to to assess patient sentiment in medical records or social media posts [4]. Now I can't sit here and list out all its potential use cases, however its uses above are just a small selection of its capabilities. This highlights why this field is of interest to many.\n",
"\n",
"I will be conducting sentiment analysis utilising the Amazon Dataset. The Amazon reviews dataset \"*consists of reviews from amazon [, the] data span[s] a period of 18 years, including ~35 million reviews up to March 2013*\"[5]. My input feature will be the review content, and my output will be the review polarity 1 or 2 (where 1 is negative and 2 is positive)- though I will be normalising this. Overall, I am trying to predict the sentiment of the amazon reviews, whether they are positive or negative. The problem I am facing is a text classification problem falling under supervised learning. More specifically it is a binary classification problem. The general purpose of this report and/or machine learning task is to explore whether unknown data that hasn't been classified can lead to good predictions."
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "XrI9NcWcYyRq",
"tags": []
},
"source": [
"## 1.2 Aim and Objectives"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "Nmu4f3lxZ-Sp",
"tags": []
},
"source": [
"### 1.2.1 Objectives"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "bliW__6Ucgpk"
},
"source": [
"- Conduct data processing.\n",
"- Write modular code to avoid repetition.\n",
"- Build a model that predicts the sentiments of unseen data.\n",
"- Evaluate the model."
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "53Vfft47aG2Y",
"tags": []
},
"source": [
"### 1.2.2 Aims"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "yFUb7PDKciEM"
},
"source": [
"<ol type=\"1\">\n",
" <li>Find the data needed to explore the objectives.</li>\n",
" <li>Define the problem.</li>\n",
" <li>Choose a measure of success.</li>\n",
" <li>Pick an evaluation protocol.</li>\n",
" <li>\n",
" Prepare the data.\n",
" <ul>\n",
" <li>Convert the textual data into numerical data.</li>\n",
" <li>Convert the numerical data into tensors.</li>\n",
" </ul>\n",
" </li>\n",
" <li>Develop a model that does better than the baseline.</li>\n",
" <li>Develop a model that overfits.</li>\n",
" <li>Regularize the model and tune the hyperparameters.</li>\n",
" <li>Evaluate the model.</li>\n",
"</ol>"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "Lwg02d_IY_Dv",
"tags": []
},
"source": [
"## 1.3 Dataset"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "PyO5pNXkykMR",
"tags": []
},
"source": [
"### 1.3.1 Limitations and dataset modifications"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "y0J5CuPzjHwe"
},
"source": [
"I initially imported my data into a jupyter notebook that I was running on my own device. As previously mentioned, the *Amazon Reviews Polarity Dataset* [5] is very large, however I had no problem importing the data from a `csv` file and manipulating it. In the jupyter notebook, I created a class (more about it later in this section) that converted my `csv` files into dataframes that could be used to train and evaluate a model. With that said, it all went pear-shaped when I decided to instead use Google Colab to run the machine learning tasks more effectively.\n",
"\n",
"The first error I ran into was `ParserError: Error tokenizing data. C error: EOF inside string`. After a lot of documentation reading and StackOverflow reviewing, I thought that the problem was due to special characters in my dataset not being properly enclosed or handled. With this information in hand, I updated the parameters I passed to the Pandas `read_csv()` method in my `csv_to_df` method from\n",
"\n",
"\n",
"```\n",
"read_csv(file,\n",
" header=None,\n",
" sep=\",\",\n",
" encoding='utf-8)\n",
"```\n",
"to\n",
"\n",
"\n",
"```\n",
"read_csv(file,\n",
" header=None,\n",
" sep=\",\",\n",
" on_bad_lines=\"skip\",\n",
" engine=\"python\",\n",
" encoding='utf-8)\n",
"```\n",
"\n",
"\n",
"This seemingly solved my problem. I was able to load the data and work with it.\n",
"\n",
"Yet, upon reviewing the dataset, I realised I had a lot of missing data and my training dataset was shorter than my testing dataset. After many tedious hours of research, it dawned on me that, if the dataset loads into jupyter notebooks without any errors, then the problem most be with Google Colab. I noticed that although the training dataset was shorter than the testing dataset, the numbers were fairly similar (when I tried using different runtimes types the numbers were similar too). Therefore, I came to the conclusion that there is a limited size that can be read by Google Colab based on the runtime type when calling the `read_csv()` method.\n",
"\n",
"In order to mitigate this issue, I split the *Amazon Reviews Polarity Dataset* into smaller `csv` files (100000 rows each). I then converted them each into a dataframe, and combined the resulting dataframes- this solved the problem."
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "VnW9Rw4sKSre"
},
"source": [
"Since my dataset was too large to load into Excel or any other software on my device, I decided to write a zsh / bash script that would split my dataset (training and testing) into smaller files (I also concluded that the resulting files would be less error prone since I wasn't doing it manually). Below is the script that I wrote:\n",
"\n",
"\n",
"\n",
"```\n",
"# I wrote all this code.\n",
"\n",
"# Splitting the given file into smaller files. Each new file will have\n",
"# 100000 rows.\n",
"split -l 100000 \"$1\".csv \"$1\"_\n",
"\n",
"# Giving the created files the .csv extension.\n",
"for file in \"$1\"_*\n",
"do\n",
" mv \"$file\" \"$file.csv\"\n",
"done\n",
"```\n",
"\n",
"With this complete, I was then able to work with my data."
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "tiXW4rzDyqBI",
"tags": []
},
"source": [
"### 1.3.2 Assembling the dataset"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "hqmCoZv-0pLr"
},
"source": [
"Since I am now working with many `csv` files. I decided to create a class to avoid repeating the same code when converting the `csv` files into dataframes, and combining the dataframes.\n",
"\n",
"I will begin by importing the pandas library to make use of its DataFrame class, and the os module to generate the file paths."
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {
"executionInfo": {
"elapsed": 550,
"status": "ok",
"timestamp": 1693228087386,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "vyewCR_yy-SQ",
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"# Importing pandas.\n",
"import pandas as pd\n",
"import os"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "yKic-gKDViaf",
"tags": []
},
"source": [
"#### 1.3.2.1 `FileToDataFrame` class"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "jOuGFmmixk-r"
},
"source": [
"Next I will create a class with some of the following methods:\n",
"- `csv_to_df(file, cols)`\n",
" - This method will convert the `csv` files into dataframes, like the name suggests.\n",
"- `combine_df(file_paths, cols)`\n",
" - This method will combine the dataframes created by the `csv_to_df()` method.\n",
"- `get_file_paths(file_dir)`\n",
" - Considering that the training data was 3,600,000 lines before it was split into files 100,000 lines long resulting in 36 training data files- I need to create a method that will generate the file paths/names. This will prevent repetition and lessen the chance of errors like forgetting a file."
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {
"executionInfo": {
"elapsed": 554,
"status": "ok",
"timestamp": 1693228093277,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "goQLTqy5vU8s",
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"class FileToDataFrame:\n",
" \"\"\"\n",
" A class that generates file paths and creates dataframes from csv\n",
" files.\n",
"\n",
" Methods\n",
" -------\n",
" print_dir()\n",
" get_cols()\n",
" get_file_paths(file_dir)\n",
" csv_to_df(file, cols)\n",
" combine_df(file_paths, cols)\n",
" generate_df_from_dir()\n",
" \"\"\"\n",
"\n",
" def __init__(self, file_dir: str = None, cols: str = None):\n",
" \"\"\"\n",
" Initialises the FileToDataFrame class and sets the file directory\n",
" and column names.\n",
"\n",
" Parameters\n",
" ----------\n",
" file_dir : str, default=None\n",
" The path to the folder.\n",
" cols : list, default=None\n",
" The list of column names used to create the dataframe.\n",
" \"\"\"\n",
" self.file_dir = file_dir\n",
" self.cols = cols\n",
"\n",
" def print_dir(self):\n",
" \"\"\"\n",
" Prints the file path to the directory.\n",
"\n",
" Notes\n",
" -----\n",
" Useful for checking the directory that was initialised.\n",
" \"\"\"\n",
" print(self.file_dir)\n",
"\n",
" def get_cols(self):\n",
" \"\"\"\n",
" Returns the column names providing during initialisation.\n",
"\n",
" Returns\n",
" -------\n",
" cols : list\n",
" Returns the list of column names used to create the dataframe.\n",
" \"\"\"\n",
" return self.cols\n",
"\n",
" def get_file_paths(self, file_dir: str = None):\n",
" \"\"\"\n",
" Generates a list of file paths for a given directory.\n",
"\n",
" Parameters\n",
" ----------\n",
" file_dir : str, default=None\n",
" The path to the folder.\n",
"\n",
" Returns\n",
" -------\n",
" files : list\n",
" The list containing the filepaths.\n",
" \"\"\"\n",
" if file_dir == None:\n",
" file_dir = self.file_dir\n",
"\n",
" files = []\n",
"\n",
" # Adding the files.\n",
" for file_path in os.listdir(file_dir):\n",
" file_path_str = file_dir + \"/\" + file_path\n",
" files.append(file_path_str)\n",
"\n",
" return files\n",
"\n",
" def csv_to_df(self, file: str, cols: list):\n",
" \"\"\"\n",
" Creates a dataframe from a given file.\n",
"\n",
" Parameters\n",
" ----------\n",
" file : str\n",
" The path to the file.\n",
" cols : list\n",
" A list of the column names.\n",
"\n",
" Returns\n",
" -------\n",
" df : Pandas dataframe\n",
" The dataframe created from the file data.\n",
" \"\"\"\n",
" # Importing the data. (Setting the engine to python to run on google colabs).\n",
" csv_file = pd.read_csv(file, header=None, sep=\",\", on_bad_lines=\"skip\",\n",
" engine=\"python\", encoding='utf-8')\n",
"\n",
" # Creating an empty dictionary.\n",
" data = dict()\n",
"\n",
" # Creating a list from the data column values and appending it to the\n",
" # dictionary.\n",
" for i in range(len(cols)):\n",
" data[cols[i]] = [val for val in csv_file[csv_file.columns[i]]]\n",
"\n",
" # Creating a dataframe from the created dictionary.\n",
" df = pd.DataFrame(data)\n",
"\n",
" return df\n",
"\n",
" def combine_df(self, file_paths: list, cols: list):\n",
" \"\"\"\"\n",
" Creates and combines dataframes.\n",
"\n",
" Parameters\n",
" ----------\n",
" file_paths : list\n",
" The path to the files to convert to dataframes.\n",
" cols : list\n",
" The list of column names used to create the dataframe.\n",
"\n",
" Returns\n",
" -------\n",
" df : Pandas dataframe\n",
" The dataframe created from the combined dataframes.\n",
" \"\"\"\n",
" # Creating an empty dataframe.\n",
" df = pd.DataFrame(columns=cols)\n",
"\n",
" # Creating a new datframe for each file and appending it to df dataframe.\n",
" for file in file_paths:\n",
" df_new = self.csv_to_df(file, cols)\n",
" df = pd.concat([df, df_new], ignore_index=True)\n",
"\n",
" return df\n",
"\n",
" def generate_df_from_dir(self):\n",
" \"\"\"\"\n",
" Generates the dataframe from a given directory.\n",
"\n",
" Returns\n",
" -------\n",
" df : Pandas dataframe\n",
" The dataframe created from the files in the given directory.\n",
" \"\"\"\n",
" files = self.get_file_paths()\n",
" df = self.combine_df(files, self.cols)\n",
" return df"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "GtU0PQLrVue-",
"tags": []
},
"source": [
"#### 1.3.2.2 Creating the dataframes"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "R86XJFcJdyVq"
},
"source": [
"With the `FileToDataFrame` class created, I can now create the dataframes.\n",
"\n",
"I will begin with the training dataframe."
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"executionInfo": {
"elapsed": 208,
"status": "ok",
"timestamp": 1693228127444,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "bsn5O_kjbUcU",
"outputId": "50c9ff58-d003-496b-f90d-ee416a899240",
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"There are 36 training data files: True\n",
"./data/training\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Training data directory.\n",
"training_dir = \"./data/training\"\n",
"\n",
"# Initialising the FileToDataFrame class\n",
"amazon_train = FileToDataFrame(training_dir,\n",
" [\"polarity\", \"review_title\", \"review_content\"])\n",
"\n",
"# Checking that there are 36 files.\n",
"print(f\"There are 36 training data files: {len(amazon_train.get_file_paths()) == 36}\")\n",
"\n",
"# Checking the directory is correct.\n",
"amazon_train.print_dir()"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "JYnISQ8u4Crm"
},
"source": [
"Now that I have verified that there are 36 training data files I can generate the dataframe."
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {
"executionInfo": {
"elapsed": 40340,
"status": "ok",
"timestamp": 1693228171215,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "OLaiG3Rv3Qb5",
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"# Generating the training dataframe.\n",
"train_data = amazon_train.generate_df_from_dir()"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/",
"height": 424
},
"executionInfo": {
"elapsed": 220,
"status": "ok",
"timestamp": 1693228175144,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "ZnwOPKz93tr-",
"outputId": "7aceaf3f-fea5-4527-e0d7-92c9b530a3a8",
"tags": []
},
"outputs": [
{
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>polarity</th>\n",
" <th>review_title</th>\n",
" <th>review_content</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>0</th>\n",
" <td>1</td>\n",
" <td>More posturing than substance</td>\n",
" <td>The first thing anyone who reads this book nee...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1</th>\n",
" <td>1</td>\n",
" <td>The Courage of Captain Plum</td>\n",
" <td>James Oliver Curwood wrote many wonderful stor...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2</th>\n",
" <td>2</td>\n",
" <td>the Courage of Captain Plum</td>\n",
" <td>Kind of hoped for a hardback, but guess that s...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3</th>\n",
" <td>2</td>\n",
" <td>Great movie</td>\n",
" <td>Great Jimmy Stewart movie. A very conservative...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <td>2</td>\n",
" <td>The FBI Story</td>\n",
" <td>We both really enjoy catching up watching many...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>...</th>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3599995</th>\n",
" <td>2</td>\n",
" <td>Not as good as expected</td>\n",
" <td>First off, let me say this is a great lens. It...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3599996</th>\n",
" <td>2</td>\n",
" <td>Quality lens</td>\n",
" <td>Very good quality lens, feels solid and the pi...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3599997</th>\n",
" <td>2</td>\n",
" <td>Haven't used it enough.</td>\n",
" <td>Haven't used it enough to make an informed dec...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3599998</th>\n",
" <td>2</td>\n",
" <td>Canon L Lens at its Best</td>\n",
" <td>Got this lens with my Canon 5D Mark II and it ...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3599999</th>\n",
" <td>2</td>\n",
" <td>Great lens</td>\n",
" <td>I've had this lens for 2 months now, and love ...</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"<p>3600000 rows × 3 columns</p>\n",
"</div>"
],
"text/plain": [
" polarity review_title \\\n",
"0 1 More posturing than substance \n",
"1 1 The Courage of Captain Plum \n",
"2 2 the Courage of Captain Plum \n",
"3 2 Great movie \n",
"4 2 The FBI Story \n",
"... ... ... \n",
"3599995 2 Not as good as expected \n",
"3599996 2 Quality lens \n",
"3599997 2 Haven't used it enough. \n",
"3599998 2 Canon L Lens at its Best \n",
"3599999 2 Great lens \n",
"\n",
" review_content \n",
"0 The first thing anyone who reads this book nee... \n",
"1 James Oliver Curwood wrote many wonderful stor... \n",
"2 Kind of hoped for a hardback, but guess that s... \n",
"3 Great Jimmy Stewart movie. A very conservative... \n",
"4 We both really enjoy catching up watching many... \n",
"... ... \n",
"3599995 First off, let me say this is a great lens. It... \n",
"3599996 Very good quality lens, feels solid and the pi... \n",
"3599997 Haven't used it enough to make an informed dec... \n",
"3599998 Got this lens with my Canon 5D Mark II and it ... \n",
"3599999 I've had this lens for 2 months now, and love ... \n",
"\n",
"[3600000 rows x 3 columns]"
]
},
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Viewing the created dataframe.\n",
"train_data"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "Jv1-plDG5VnG"
},
"source": [
"Now I will create the testing dataframe."
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"executionInfo": {
"elapsed": 223,
"status": "ok",
"timestamp": 1693228180268,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "lu_CEVl6zhcR",
"outputId": "979727fa-b222-4886-c414-512e54b817c5",
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"There are 4 testing data files: True\n",
"./data/testing\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Testing data directory.\n",
"testing_dir = \"./data/testing\"\n",
"\n",
"# Initialising the FileToDataFrame class\n",
"amazon_test = FileToDataFrame(testing_dir,\n",
" [\"polarity\", \"review_title\", \"review_content\"])\n",
"\n",
"# Checking that there are 4 files.\n",
"print(f\"There are 4 testing data files: {len(amazon_test.get_file_paths()) == 4}\")\n",
"\n",
"# Checking the directory is correct.\n",
"amazon_test.print_dir()"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "141f76QC4IVR"
},
"source": [
"Now that I have verified that there are 4 testing data files I can generate the dataframe."
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {
"executionInfo": {
"elapsed": 4035,
"status": "ok",
"timestamp": 1693228187327,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "zLdU8XAh3nyI",
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"# Generating the testing dataframe.\n",
"test_data = amazon_test.generate_df_from_dir()"
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/",
"height": 424
},
"executionInfo": {
"elapsed": 221,
"status": "ok",
"timestamp": 1693228189787,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "EehmFo_m36wn",
"outputId": "ebfef6bc-f24b-44c8-a41a-7edc86be7c40",
"tags": []
},
"outputs": [
{
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>polarity</th>\n",
" <th>review_title</th>\n",
" <th>review_content</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>0</th>\n",
" <td>2</td>\n",
" <td>Useful for remodels</td>\n",
" <td>I recently remodeled my house and these came i...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1</th>\n",
" <td>2</td>\n",
" <td>decent get what u pay for</td>\n",
" <td>I got this set for around the house and for th...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2</th>\n",
" <td>2</td>\n",
" <td>Rock solid</td>\n",
" <td>Perfect solution - stable and accessible. Perh...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3</th>\n",
" <td>2</td>\n",
" <td>Fun, humorous, and touching!</td>\n",
" <td>I highly recommend this movie ... it's a beaut...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <td>2</td>\n",
" <td>Horror that isn't for the faint-hearted</td>\n",
" <td>Of the fourteen books of Shaun Hutson's that I...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>...</th>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>399995</th>\n",
" <td>1</td>\n",
" <td>Inferior product and not like the picture</td>\n",
" <td>I expected the 2-piece filter set shown in the...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>399996</th>\n",
" <td>1</td>\n",
" <td>Entire Set Was Not Recieved</td>\n",
" <td>I rec'd the filter, but not the plastic part t...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>399997</th>\n",
" <td>1</td>\n",
" <td>No Darn Good!</td>\n",
" <td>Now that we know this is an interview CD, I am...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>399998</th>\n",
" <td>2</td>\n",
" <td>Revenge</td>\n",
" <td>This is really just a typical revenge movie wi...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>399999</th>\n",
" <td>1</td>\n",
" <td>book was somewhat interesting but too negative</td>\n",
" <td>The author took a very negative approach to re...</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"<p>400000 rows × 3 columns</p>\n",
"</div>"
],
"text/plain": [
" polarity review_title \\\n",
"0 2 Useful for remodels \n",
"1 2 decent get what u pay for \n",
"2 2 Rock solid \n",
"3 2 Fun, humorous, and touching! \n",
"4 2 Horror that isn't for the faint-hearted \n",
"... ... ... \n",
"399995 1 Inferior product and not like the picture \n",
"399996 1 Entire Set Was Not Recieved \n",
"399997 1 No Darn Good! \n",
"399998 2 Revenge \n",
"399999 1 book was somewhat interesting but too negative \n",
"\n",
" review_content \n",
"0 I recently remodeled my house and these came i... \n",
"1 I got this set for around the house and for th... \n",
"2 Perfect solution - stable and accessible. Perh... \n",
"3 I highly recommend this movie ... it's a beaut... \n",
"4 Of the fourteen books of Shaun Hutson's that I... \n",
"... ... \n",
"399995 I expected the 2-piece filter set shown in the... \n",
"399996 I rec'd the filter, but not the plastic part t... \n",
"399997 Now that we know this is an interview CD, I am... \n",
"399998 This is really just a typical revenge movie wi... \n",
"399999 The author took a very negative approach to re... \n",
"\n",
"[400000 rows x 3 columns]"
]
},
"execution_count": 8,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Viewing the created dataframe.\n",
"test_data"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "yKaWSMYW5o6U"
},
"source": [
"Now that I have created the training and testing dataframes I will check the size and shape of them, to verify that no errors occured."
]
},
{
"cell_type": "code",
"execution_count": 9,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"executionInfo": {
"elapsed": 207,
"status": "ok",
"timestamp": 1693228193063,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "y2WvsLASEprT",
"outputId": "0c17a2c5-e99d-4cb7-ed92-2d4ed1ff3a80",
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"DATA SIZE\n",
"---------\n",
"Training Data: 3600000 \t(3600000, 3)\n",
"Testing Data : 400000 \t(400000, 3)\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Printing the size and shape of the data.\n",
"print(\"DATA SIZE\")\n",
"print(\"---------\")\n",
"print(f\"Training Data: {len(train_data)} \\t{train_data.shape}\")\n",
"print(f\"Testing Data : {len(test_data)} \\t{test_data.shape}\")"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 1.3.3 Reducing the size of the dataset."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"I originally reached section 5 and was about to begin building models, but was forced to come back to this section and reduce the amount of data, since my notebook environment kept on crashing, disenabling me from going further in the DLWP process."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The ratio of positive reviews to negative ones was equal before the split, so it was important to keep it that way. If it changed, I would be dealing with a completely different problem."
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Negative Reviews : 1800000\n",
"Positive Reviews : 1800000\n",
"Reviews are equal: True\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Checking the ratio of 1 (negative) or 2 (positive) (to check whether it is\n",
"# balanced).\n",
"num_neg_reviews = len(train_data.loc[train_data[\"polarity\"] == 1])\n",
"num_pos_reviews = len(train_data.loc[train_data[\"polarity\"] == 2])\n",
"reviews_is_equal = (num_neg_reviews == num_pos_reviews)\n",
"\n",
"print(f\"Negative Reviews : {num_neg_reviews}\")\n",
"print(f\"Positive Reviews : {num_pos_reviews}\")\n",
"print(f\"Reviews are equal: {reviews_is_equal}\")"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"#### 1.3.3.1 `split_dataset` function"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The function below returns a fraction of the given dataset, in this case I want half of the dataset. It takes into the account the ratio of positive to negative reviews."
]
},
{
"cell_type": "code",
"execution_count": 11,
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"def split_dataset(df, frac: float = 0.5, col: str = \"polarity\"):\n",
" \"\"\"\n",
" A function that returns a fraction of the given dataset.\n",
"\n",
" Parameters\n",
" ----------\n",
" df : Pandas DataFrame\n",
" The dataset that the sample will be taken from.\n",
" frac: float, default=0.5\n",
" The fraction of the dataset to sample.\n",
" col: str, default=\"polarity\"\n",
" Label column of the dataset\n",
"\n",
" Returns\n",
" -------\n",
" _ : Pandas DataFrame\n",
" Returns a sample of the Pandas DataFrame.\n",
" \"\"\"\n",
" # Checking that the inputted DataFrame is not None or empty.\n",
" try:\n",
" if df is None or df.empty:\n",
" raise ValueError(\"The inputted DataFrame is None or empty.\")\n",
"\n",
" # Separating the data into positive (2) and negative (1) data.\n",
" pos_data = df[df[col] == 2]\n",
" neg_data = df[df[col] == 1]\n",
"\n",
" # Calculating the amount of data to take from each category.\n",
" num_pos_data = int(frac * len(pos_data))\n",
" num_neg_data = int(frac * len(neg_data))\n",
"\n",
" # Splitting / sampling the dataset.\n",
" train_pos = pos_data.sample(n=num_pos_data, random_state=1)\n",
" train_neg = neg_data.sample(n=num_neg_data, random_state=1)\n",
"\n",
" #  Combining the positive and negative datasets.\n",
" df = pd.concat([train_pos, train_neg], ignore_index=True)\n",
"\n",
" # Returning the combined dataset.\n",
" return df\n",
" except Exception as e:\n",
" print(f\"ERROR OCCURRED: {e}\")\n",
" return None"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"#### 1.3.3.2 Halving the training and testing datasets."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"I will now reduce the size of the data with the created function."
]
},
{
"cell_type": "code",
"execution_count": 12,
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell\n",
"# Reducing the size of data using the split_dataset function.\n",
"half_train_data = split_dataset(train_data)\n",
"half_test_data = split_dataset(test_data)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Viewing the new training dataset."
]
},
{
"cell_type": "code",
"execution_count": 13,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>polarity</th>\n",
" <th>review_title</th>\n",
" <th>review_content</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>0</th>\n",
" <td>2</td>\n",
" <td>Nice Hat</td>\n",
" <td>You can't beat Henschel for well made and reas...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1</th>\n",
" <td>2</td>\n",
" <td>Jesus Reborn</td>\n",
" <td>I once walked the earth in sin. I was a non-be...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2</th>\n",
" <td>2</td>\n",
" <td>World's Toughest Computer</td>\n",
" <td>I've been saying for a month that I should wri...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3</th>\n",
" <td>2</td>\n",
" <td>An informative book about figure skating...</td>\n",
" <td>Yamaguchi's Figure Skating for Dummies is an e...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <td>2</td>\n",
" <td>Viva Las Vegas!!!!!!</td>\n",
" <td>I had never heard of Dread Zeppelin until two ...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>...</th>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1799995</th>\n",
" <td>1</td>\n",
" <td>not for kids</td>\n",
" <td>The poses are too advanced for younger childre...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1799996</th>\n",
" <td>1</td>\n",
" <td>a note on longevity</td>\n",
" <td>prior versions were impregnated with iodine an...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1799997</th>\n",
" <td>1</td>\n",
" <td>FYI Made in China</td>\n",
" <td>Pretty pricey considering it's MADE IN CHINA. ...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1799998</th>\n",
" <td>1</td>\n",
" <td>Breaking bits and dead drills</td>\n",
" <td>What's the saying? Three strikes, you're out? ...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1799999</th>\n",
" <td>1</td>\n",
" <td>Great Sphinx puzzle</td>\n",
" <td>Too hard for our house- the colors are very si...</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"<p>1800000 rows × 3 columns</p>\n",
"</div>"
],
"text/plain": [
" polarity review_title \\\n",
"0 2 Nice Hat \n",
"1 2 Jesus Reborn \n",
"2 2 World's Toughest Computer \n",
"3 2 An informative book about figure skating... \n",
"4 2 Viva Las Vegas!!!!!! \n",
"... ... ... \n",
"1799995 1 not for kids \n",
"1799996 1 a note on longevity \n",
"1799997 1 FYI Made in China \n",
"1799998 1 Breaking bits and dead drills \n",
"1799999 1 Great Sphinx puzzle \n",
"\n",
" review_content \n",
"0 You can't beat Henschel for well made and reas... \n",
"1 I once walked the earth in sin. I was a non-be... \n",
"2 I've been saying for a month that I should wri... \n",
"3 Yamaguchi's Figure Skating for Dummies is an e... \n",
"4 I had never heard of Dread Zeppelin until two ... \n",
"... ... \n",
"1799995 The poses are too advanced for younger childre... \n",
"1799996 prior versions were impregnated with iodine an... \n",
"1799997 Pretty pricey considering it's MADE IN CHINA. ... \n",
"1799998 What's the saying? Three strikes, you're out? ... \n",
"1799999 Too hard for our house- the colors are very si... \n",
"\n",
"[1800000 rows x 3 columns]"
]
},
"execution_count": 13,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Viewing the reduced dataframe.\n",
"half_train_data"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Viewing the new testing dataset."
]
},
{
"cell_type": "code",
"execution_count": 14,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>polarity</th>\n",
" <th>review_title</th>\n",
" <th>review_content</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>0</th>\n",
" <td>2</td>\n",
" <td>So cute</td>\n",
" <td>My son loves it. My complaint is that it laste...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1</th>\n",
" <td>2</td>\n",
" <td>Great movie</td>\n",
" <td>I am an absolute fan of the Crow series. I own...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2</th>\n",
" <td>2</td>\n",
" <td>Best Baby Einstein yet!</td>\n",
" <td>Baby Noah is the best Baby Einstein for any ag...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3</th>\n",
" <td>2</td>\n",
" <td>Member of the pack</td>\n",
" <td>Fine story with well drawn characters and terr...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <td>2</td>\n",
" <td>Love this</td>\n",
" <td>My friend has this album for her daughter n I ...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>...</th>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>199995</th>\n",
" <td>1</td>\n",
" <td>Love VNV But Hate This DVD.</td>\n",
" <td>I practically like every track they have ever ...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>199996</th>\n",
" <td>1</td>\n",
" <td>don't buy</td>\n",
" <td>We needed one and this was what was available ...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>199997</th>\n",
" <td>1</td>\n",
" <td>did not work for long</td>\n",
" <td>We bought 2 of these to use with a remote cont...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>199998</th>\n",
" <td>1</td>\n",
" <td>Not what I expected at all</td>\n",
" <td>I purchased this product under the assumption ...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>199999</th>\n",
" <td>1</td>\n",
" <td>Flexrake 100A</td>\n",
" <td>I have had a hula hoe for year. It has worked ...</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"<p>200000 rows × 3 columns</p>\n",
"</div>"
],
"text/plain": [
" polarity review_title \\\n",
"0 2 So cute \n",
"1 2 Great movie \n",
"2 2 Best Baby Einstein yet! \n",
"3 2 Member of the pack \n",
"4 2 Love this \n",
"... ... ... \n",
"199995 1 Love VNV But Hate This DVD. \n",
"199996 1 don't buy \n",
"199997 1 did not work for long \n",
"199998 1 Not what I expected at all \n",
"199999 1 Flexrake 100A \n",
"\n",
" review_content \n",
"0 My son loves it. My complaint is that it laste... \n",
"1 I am an absolute fan of the Crow series. I own... \n",
"2 Baby Noah is the best Baby Einstein for any ag... \n",
"3 Fine story with well drawn characters and terr... \n",
"4 My friend has this album for her daughter n I ... \n",
"... ... \n",
"199995 I practically like every track they have ever ... \n",
"199996 We needed one and this was what was available ... \n",
"199997 We bought 2 of these to use with a remote cont... \n",
"199998 I purchased this product under the assumption ... \n",
"199999 I have had a hula hoe for year. It has worked ... \n",
"\n",
"[200000 rows x 3 columns]"
]
},
"execution_count": 14,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Viewing the reduced dataframe.\n",
"half_test_data"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now I will print the new sizes of the datasets."
]
},
{
"cell_type": "code",
"execution_count": 15,
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"DATA SIZE\n",
"---------\n",
"Training Data: 1800000 \t(1800000, 3)\n",
"Testing Data : 200000 \t(200000, 3)\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Printing the size and shape of the data.\n",
"print(\"DATA SIZE\")\n",
"print(\"---------\")\n",
"print(f\"Training Data: {len(half_train_data)} \\t{half_train_data.shape}\")\n",
"print(f\"Testing Data : {len(half_test_data)} \\t{half_test_data.shape}\")"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "LNWy8jBsZJ0H",
"tags": []
},
"source": [
"## 1.4 Constraints and Ethical Considerations"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Since I am using a dataset that consists of the reviews of many different people there are some things to consider. The first is the privacy concern, the dataset shouldn't contain any personal information. In this case, the dataset I am using has already been processed to eliminate all private information. I must also be transparent and clearly document how the data was collected. Again, this has already been done, and the link to the dataset I used is in my references. There are other ethical concerns but most of them have been already been mitigated, I just have to understand the implications of my report / ML task.\n",
"\n",
"One constraint to mention and that has decided to haunt me is the data size. Large datasets require substantial storage and processing power. The Amazon dataset is very large, and I have already started to see the drawbacks of using such a large dataset. My limited computing resources has caused some limitations, which will be seen later on in the report / ML task."
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "NRCzbf3dIX7w",
"tags": []
},
"source": [
"# 2 Choosing a measure of success"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "EDbgF7ry94ef"
},
"source": [
"Now that I have the datasets ready, I need to decide on a measure of success. In order to do so, I will take a closer look at the training data.\n",
"\n",
"First, I will begin by checking the polarity column, since it is holds the target / labels data.\n",
"\n",
"I am under the assumption that there are only two unique values, 1 and 2. 1 representing negative reviews and 2 representing positive reviews. However when I was reading up on how the dataset was constructed I found that initially 1 and 2 were negative, 4 and 5 were positive, and 3 was ignored. Therefore, I have to check whether that data that I acquired has only 1 and 2 as values. If it includes 4 and 5, I will then have to further process the dataframes."
]
},
{
"cell_type": "code",
"execution_count": 16,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"executionInfo": {
"elapsed": 212,
"status": "ok",
"timestamp": 1693228204731,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "l_lwvVo7Kc9p",
"outputId": "2e21330a-a8f0-4c7e-89ba-86063b434e2e",
"tags": []
},
"outputs": [
{
"data": {
"text/plain": [
"{1, 2}"
]
},
"execution_count": 16,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Checking that polarity can only be 1 (negative) or 2 (positive).\n",
"set(half_train_data[\"polarity\"])"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "P2DkaOrbCUyT"
},
"source": [
"As it turns out, there are only 1s and 2s, therefore I will only need to vectorize the labels later.\n",
"\n",
"Next I will check the ratio of 1s to 2s since this will influence the measure of success that I will select."
]
},
{
"cell_type": "code",
"execution_count": 17,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"executionInfo": {
"elapsed": 1217,
"status": "ok",
"timestamp": 1693228207566,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "hHYzCGOv4M5u",
"outputId": "f654f21a-a1eb-4f94-a02b-2915ed5919e5",
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Negative Reviews : 900000\n",
"Positive Reviews : 900000\n",
"Reviews are equal: True\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Checking the ratio of 1 (negative) or 2 (positive) (to check whether it is\n",
"# balanced).\n",
"num_neg_reviews = len(half_train_data.loc[half_train_data[\"polarity\"] == 1])\n",
"num_pos_reviews = len(half_train_data.loc[half_train_data[\"polarity\"] == 2])\n",
"reviews_is_equal = (num_neg_reviews == num_pos_reviews)\n",
"\n",
"print(f\"Negative Reviews : {num_neg_reviews}\")\n",
"print(f\"Positive Reviews : {num_pos_reviews}\")\n",
"print(f\"Reviews are equal: {reviews_is_equal}\")"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "QB6gInE1HLb2"
},
"source": [
"Since the ratio of negative reviews to positive reviews is equal (there is not an inbalance), my measure of success will be accuracy rather than precision or recall. This is because accuracy is best suited for when classes are balanced. In addition, accuracy is simple and will measure the percentage of correctly predicted polarities- which is what I am aiming for.\n",
"\n",
"On the other hand, precision and recall are more suitable for datasets with class imbalances, since precision measures the true positive predictions compared to all positive predictions, and recall measures the true positive predictions compared to all actual positives.\n",
"\n",
"I will not be using ROC AUC (area under the receiver operating characteristic curve) as it better suited for evaluating the model's ability to distinguish between the classes across different thresholds."
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "4X1a_Tt1KekO",
"tags": []
},
"source": [
"# 3 Deciding on an evaluation protocol"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "TbM0djY3K7Gs"
},
"source": [
"Since I have decided on a measure of success, it is now time to pick an evaluation protocol. As mentioned before, my dataset was pretty large 3.6 million lines (see *section 1.3.2.2*), 4 million if I included the test dataset. However, the size of the dataset was leading to numerous problems, therefore I decided to use half of the dataset. Now, my dataset is 1.8 million lines, 2 million if I include the reduced test dataset. This is still very large. Since the dataset is quite large I have the freedom to pick any (to certain degree) evaluation protocol that I would like to use.\n",
"\n",
"Due to this, I will be utilising the hold-out validation set approach since it requires less computational resources compared to k-fold cross validation and iterated k-fold validation, which in turn makes it faster. It works very well with larger datasets, and is great for model complexity tuning.\n",
"\n",
"I will not be using k-fold cross validation or iterated k-fold validation since they better suit smaller datasets. K-fold cross validation is great when you have too \"*few samples for hold out validation to be reliable*\"[6]. Iterated k-fold validation is great for performing \"*highly accurate model evaluation when little data is available*\"[6]."
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "T-gceqrbLV3O",
"tags": []
},
"source": [
"# 4 Preparing the data"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now that I know what I am training, what I am optimising for, and how to evaluate my approach, I need to format my data, so that it can be fed into a machine learning model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"I will begin by importing the relevant libraries."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"executionInfo": {
"elapsed": 3014,
"status": "ok",
"timestamp": 1693228217535,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "oqcrgFAp6p5H",
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"# Importing the relevant libraries.\n",
"import tensorflow as tf\n",
"from tensorflow import keras\n",
"from keras.preprocessing.text import Tokenizer\n",
"import numpy as np\n",
"from keras import models\n",
"from keras import layers\n",
"from keras import optimizers\n",
"from keras import losses\n",
"from keras import metrics\n",
"from keras.optimizers import RMSprop\n",
"import matplotlib.pyplot as plt"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"## 4.1 The `VectorizeDataset` class"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"I decided to create a class called `VectorizeDataset` to simplify the vectorisation process. The class contains the following methods:\n",
"1. `vectorize_labels(isTest)`\n",
"2. `vectorize_inputs(isTest)`\n",
"\n",
"Like the name suggests the`vectorize_labels()` method converts the given labels into their corresponding one-hot binary representations, whereas the `vectorize_inputs()` method converts the given inputs (data / features) into numerical representations that can be used as inputs to a neural network."
]
},
{
"cell_type": "code",
"execution_count": 19,
"metadata": {
"executionInfo": {
"elapsed": 195,
"status": "ok",
"timestamp": 1693228219742,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "63zB2HrLMa9B",
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"class VectorizeDataset:\n",
" \"\"\"\n",
" A class that vectorises the given dataset.\n",
"\n",
" Methods\n",
" -------\n",
" vectorize_labels(isTest)\n",
" vectorize_inputs(isTest)\n",
" \"\"\"\n",
"\n",
" def __init__(self, dataset_train, dataset_test, labels: str, inputs: str,\n",
" max_words: int = 10000):\n",
" \"\"\"\n",
" Initialises the VectorizeDataset class and sets the dataset,\n",
" labels, and inputs.\n",
"\n",
" Parameters\n",
" ----------\n",
" dataset_train : Pandas.DataFrame\n",
" The training dataset from which the labels and inputs will be extracted.\n",
" dataset_test : Pandas.DataFrame\n",
" The testing dataset from which the labels and inputs will be extracted.\n",
" labels : str\n",
" The column name that refers to the target / output data.\n",
" inputs : str\n",
" The column name that refers to the feature / input data.\n",
" max_words : int, default=10000\n",
" The maximum vocabulary size.\n",
" \"\"\"\n",
" self.dataset_train = dataset_train\n",
" self.dataset_test = dataset_test\n",
" self.labels = labels\n",
" self.inputs = inputs\n",
" self.max_words = max_words\n",
"\n",
" def vectorize_labels(self, isTest: bool = False):\n",
" \"\"\"\n",
" Converts the given labels into their corresponding one-hot binary\n",
" representations.\n",
"\n",
" Parameters\n",
" ----------\n",
" isTest : bool, default=False\n",
" Checks whether the data to vectorise is testing data or training data.\n",
"\n",
" Returns\n",
" -------\n",
" _ : tensor\n",
" Returns a tensor containing the one-hot binary representations.\n",
" \"\"\"\n",
" # Extracting the labels column data.\n",
" labels_col = self.dataset_train[self.labels].values if (\n",
" isTest == False) else self.dataset_test[self.labels].values\n",
" \n",
" # Normalising the labels by subtracting 1.\n",
" return labels_col - 1\n",
"\n",
" def vectorize_inputs(self, isTest: bool = False):\n",
" \"\"\"\n",
" Converts the given inputs (data / features) into numerical\n",
" representations that can be used as inputs to a neural network.\n",
"\n",
" Parameters\n",
" ----------\n",
" isTest : bool, default=False\n",
" Checks whether the data to vectorise is testing data or training data.\n",
"\n",
" Returns\n",
" -------\n",
" one_hot_data : tensor\n",
" Returns a tensor holding the numerical representation of the input\n",
" data.\n",
" \"\"\"\n",
" # Extracting the inputs column data.\n",
" inputs_col = self.dataset_train[self.inputs].values\n",
"\n",
" # Initialising the Tokenizer + setting max_words to limit vocabulary size.\n",
" tokenizer = Tokenizer(num_words=self.max_words)\n",
"\n",
" # Fitting the tokenizer to the given inputs\n",
" tokenizer.fit_on_texts(inputs_col)\n",
"\n",
" if (isTest == False):\n",
" # Converting the inputs into sequences of integers.\n",
" sequences = tokenizer.texts_to_sequences(inputs_col)\n",
"\n",
" # Performing one-hot encoding on the created sequences.\n",
" one_hot_data = tokenizer.sequences_to_matrix(\n",
" sequences, mode='binary')\n",
" else:\n",
" # Extracting the inputs test column data.\n",
" inputs_test_col = self.dataset_test[self.inputs].values\n",
"\n",
" # Converting the test inputs into sequences of integers.\n",
" sequences = tokenizer.texts_to_sequences(inputs_test_col)\n",
"\n",
" # Performing one-hot encoding on the created sequences.\n",
" one_hot_data = tokenizer.sequences_to_matrix(\n",
" sequences, mode='binary')\n",
"\n",
" return one_hot_data"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"## 4.2 Vectorising the data"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "Ye0BexDFCiUZ"
},
"source": [
"I initially planned on setting `max_words` (which is the maximum vocabulary size) to 10000. However, my notebook environment kept on crashing, therefore I reduced `max_words` from 10000 to 1000. This will evidently reduce the accuracy and predictive power of my model since the vocabulary size is 10 times smaller, however since my dataset is so large, it may be able to mitigate this to a certain extent, since the model has access to a lot of training data."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"I will begin by initialising the `VectorizeDataset` class that I created, by feeding in the reduced dataset and the necessary parameters."
]
},
{
"cell_type": "code",
"execution_count": 20,
"metadata": {
"executionInfo": {
"elapsed": 214,
"status": "ok",
"timestamp": 1693228223240,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "AfzlvVCB5CpX",
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"# Initialising the VectorizeDataset class for the data.\n",
"vectorize_data = VectorizeDataset(half_train_data, half_test_data, \"polarity\",\n",
" \"review_content\", max_words=1000)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Next I will generate and view the vectorised training labels."
]
},
{
"cell_type": "code",
"execution_count": 21,
"metadata": {
"executionInfo": {
"elapsed": 389,
"status": "ok",
"timestamp": 1693228225839,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "uXHELACY5uWY",
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"# Generating the vectorised training labels.\n",
"vectorised_train_labels = vectorize_data.vectorize_labels()"
]
},
{
"cell_type": "code",
"execution_count": 22,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"executionInfo": {
"elapsed": 199,
"status": "ok",
"timestamp": 1693228228091,
"user": {
"displayName": "Worthy",
"userId": "06377054689661600867"
},
"user_tz": -60
},
"id": "aNnf17L76A-a",
"outputId": "d20380b2-e9c5-458e-cdd4-6aa92a48e619",
"tags": []
},
"outputs": [
{
"data": {
"text/plain": [
"(array([1, 1, 1, ..., 0, 0, 0], dtype=object), (1800000,))"
]
},
"execution_count": 22,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Viewing the vectorised labels and checking the shape.\n",
"vectorised_train_labels, vectorised_train_labels.shape"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now I will generate and view the vectorised training inputs."
]
},
{
"cell_type": "code",
"execution_count": 23,
"metadata": {
"id": "qHl1BeLo6OYj",
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"# Generating the vectorised training inputs.\n",
"vectorised_train_inputs = vectorize_data.vectorize_inputs()"
]
},
{
"cell_type": "code",
"execution_count": 24,
"metadata": {
"id": "ROzpq4Xz6beZ",
"tags": []
},
"outputs": [
{
"data": {
"text/plain": [
"(array([[0., 0., 1., ..., 0., 0., 0.],\n",
" [0., 1., 1., ..., 0., 0., 0.],\n",
" [0., 1., 1., ..., 0., 0., 0.],\n",
" ...,\n",
" [0., 1., 0., ..., 0., 0., 0.],\n",
" [0., 1., 1., ..., 0., 0., 0.],\n",
" [0., 1., 1., ..., 0., 0., 0.]]),\n",
" (1800000, 1000))"
]
},
"execution_count": 24,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Viewing the vectorised training inputs and checking the shape.\n",
"vectorised_train_inputs, vectorised_train_inputs.shape"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Next I will generate and view the vectorised test labels."
]
},
{
"cell_type": "code",
"execution_count": 25,
"metadata": {
"id": "DR2umDGiDGts",
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"# Generating the vectorised testing labels.\n",
"vectorised_test_labels = vectorize_data.vectorize_labels(isTest=True)"
]
},
{
"cell_type": "code",
"execution_count": 26,
"metadata": {
"id": "fxIKTZfqDPsP",
"tags": []
},
"outputs": [
{
"data": {
"text/plain": [
"(array([1, 1, 1, ..., 0, 0, 0], dtype=object), (200000,))"
]
},
"execution_count": 26,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Viewing the vectorised labels and checking the shape.\n",
"vectorised_test_labels, vectorised_test_labels.shape"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now I will generate and view the vectorised test inputs."
]
},
{
"cell_type": "code",
"execution_count": 27,
"metadata": {
"id": "KMGcruZM6kEC",
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"# Generating the vectorised testing inputs.\n",
"vectorised_test_inputs = vectorize_data.vectorize_inputs(isTest=True)"
]
},
{
"cell_type": "code",
"execution_count": 28,
"metadata": {
"id": "VYAMf0CyAOea",
"tags": []
},
"outputs": [
{
"data": {
"text/plain": [
"(array([[0., 1., 0., ..., 0., 0., 0.],\n",
" [0., 1., 0., ..., 0., 0., 0.],\n",
" [0., 1., 1., ..., 0., 0., 0.],\n",
" ...,\n",
" [0., 0., 1., ..., 0., 0., 0.],\n",
" [0., 1., 1., ..., 0., 0., 0.],\n",
" [0., 1., 1., ..., 0., 0., 0.]]),\n",
" (200000, 1000))"
]
},
"execution_count": 28,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Viewing the vectorised testing inputs and checking the shape.\n",
"vectorised_test_inputs, vectorised_test_inputs.shape"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"## 4.3 Splitting the data into training and validation to implement hold-out validation"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 4.3.1 The `BuildEvalModel` class"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Once I finished vectorising my data, I decided to create a class called `BuildEvalModel` that allows me to compile, train, and evaluate models. The class has the following methods:\n",
"1. `train_val_split(train_inputs, train_label, train_ratio)`\n",
"2. `create_compile_model(units, activation, input_shape, num_of_layers)`\n",
"3. `fit_model(model, train_x, train_y, val_x, val_y, epochs, batch_size)`\n",
"\n",
"The `train_val_split()` method splits the given data into training and validation sets. This is needed for holdout validation. The `create_compile_model()` creates and compiles a Sequential model based on the given parameters. This will stop me from manually writing out the same code. The `fit_model()` fits the model- again needed to avoid unnecessary code repetition, leading to modular code."
]
},
{
"cell_type": "code",
"execution_count": 29,
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"class BuildEvalModel:\n",
" \"\"\"\n",
" A class that compiles, trains, and evaluates a model.\n",
"\n",
" Methods\n",
" -------\n",
" train_val_split(train_inputs, train_label, train_ratio)\n",
" create_compile_model(units, activation, input_shape, num_of_layers)\n",
" fit_model(model, train_x, train_y, val_x, val_y, epochs, batch_size)\n",
"\n",
" \"\"\"\n",
"\n",
" def train_val_split(self, train_inputs: list, train_labels: list,\n",
" train_ratio: float = 0.8, random_seed=None):\n",
" \"\"\"\n",
" Splits the given data into training and validation sets.\n",
"\n",
" Parameters\n",
" ----------\n",
" train_inputs : list\n",
" The input training data.\n",
" train_labels : list\n",
" The labels for the training data.\n",
" train_ratio : float, default=0.8\n",
" The ratio used to calculating the proportion of training data\n",
" to validation data.\n",
" random_seed : int, default=None\n",
" Used to set the random seed for reproducibility. \n",
"\n",
" Returns\n",
" -------\n",
" _ : tuple\n",
" Returns a tuple containing the split data.\n",
"\n",
" Notes\n",
" -----\n",
" tuple[0] corresponds to the training data (inputs).\n",
" tuple[1] corresponds to the validation data (inputs).\n",
" tuple[2] corresponds to the training labels (targets).\n",
" tuple[3] corresponds to the validation labels (targets).\n",
" \"\"\"\n",
" # Checking whether the given data are numpy arrays.\n",
" if not isinstance(train_inputs, np.ndarray):\n",
" train_inputs = np.array(train_inputs)\n",
" if not isinstance(train_labels, np.ndarray):\n",
" train_labels = np.array(train_labels)\n",
"\n",
" # Setting the random seed for reproducibility.\n",
" if random_seed is not None:\n",
" np.random.seed(random_seed)\n",
"\n",
" # Checking that the length of the labels and inputs.\n",
" if (len(train_inputs) != len(train_labels)):\n",
" print(\"ERROR: The length of inputs doesn't equal the length of labels.\")\n",
" return\n",
"\n",
" # Calculating the number of data for the train and val data based on the train ratio.\n",
" total_data = len(train_inputs)\n",
" train_data = int(train_ratio * total_data)\n",
"\n",
" # Randomising / shuffling the data and labels.\n",
" indices = np.arange(total_data)\n",
" np.random.shuffle(indices)\n",
" train_inputs = train_inputs[indices]\n",
" train_labels = train_labels[indices]\n",
"\n",
" # Splitting the data into training and validation sets.\n",
" X_train = train_inputs[:train_data]\n",
" X_val = train_inputs[train_data:]\n",
" y_train = train_labels[:train_data]\n",
" y_val = train_labels[train_data:]\n",
"\n",
" return (X_train, X_val, y_train, y_val)\n",
"\n",
" def create_compile_model(self, units: list, activation: list,\n",
" input_shape=(1000,), num_of_layers: int = 3):\n",
" \"\"\"\n",
" Creates and compiles a Sequential model based on the given parameters.\n",
"\n",
" Parameters\n",
" ----------\n",
" units : list of integers (positive)\n",
" List containing the dimensionalities of the output space.\n",
" activation : list of str\n",
" Specifies the activation function to be used.\n",
" input_shape : tuple, default=(1000,)\n",
" The shape that corresponds to structure of the chosen data.\n",
" num_of_layers : int, default=3\n",
" The number of layers that the model will have.\n",
"\n",
" Returns\n",
" -------\n",
" model : keras Sequential object\n",
" Returns a model based on the given parameters.\n",
" \"\"\"\n",
" #  Creating the model.\n",
" model = keras.Sequential()\n",
"\n",
" # Adding models to the layers.\n",
" for i in range(num_of_layers):\n",
" if i == 0:\n",
" model.add(layers.Dense(units[i], activation=activation[i],\n",
" input_shape=input_shape))\n",
" else:\n",
" model.add(layers.Dense(\n",
" units[i],\n",
" activation=activation[i]\n",
" ))\n",
"\n",
" # Compiling the model.\n",
" model.compile(\n",
" optimizer='rmsprop',\n",
" loss='binary_crossentropy',\n",
" metrics=['accuracy']\n",
" )\n",
"\n",
" return model\n",
"\n",
" def fit_model(self, model, train_x, train_y, val_x, val_y, epochs: int = 20,\n",
" batch_size: int = 512):\n",
" \"\"\"\n",
" Fits the model.\n",
"\n",
" Parameters\n",
" ----------\n",
" model : keras.Sequential object\n",
" The model that will be built / fitted.\n",
" train_x : Array-like structure\n",
" The input data.\n",
" train_y : list / tensor / array\n",
" The target data.\n",
" val_x : list / tensor / array\n",
" The validation data.\n",
" val_y : list / array\n",
" The validation data.\n",
" epochs : int, default=20\n",
" The number of epochs to train the model.\n",
" batch_size : int, default=512\n",
" The number of samples per gradient update.\n",
"\n",
" Returns\n",
" -------\n",
" history : History object\n",
" Returns a History object where the attribute \"history\" holds a \n",
" record of training loss and other metrics.\n",
" \"\"\"\n",
" # Training the model.\n",
" history = model.fit(\n",
" train_x,\n",
" train_y,\n",
" epochs=epochs,\n",
" batch_size=batch_size,\n",
" validation_data=(val_x, val_y)\n",
" )\n",
"\n",
" return history"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 4.3.2 Converting the `vectorised_train_labels` and `vectorised_test_labels`"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Before I split the data, I need to double check that my data is of the right type."
]
},
{
"cell_type": "code",
"execution_count": 30,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"text/plain": [
"(dtype('float64'), dtype('O'))"
]
},
"execution_count": 30,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Checking the type of the training data.\n",
"vectorised_train_inputs.dtype, vectorised_train_labels.dtype"
]
},
{
"cell_type": "code",
"execution_count": 31,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"text/plain": [
"(dtype('float64'), dtype('O'))"
]
},
"execution_count": 31,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Checking the type of the testing data.\n",
"vectorised_test_inputs.dtype, vectorised_test_labels.dtype"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"Since there are a couple variables with the wrong type (`object`). I will create a function to convert the type (to `float64`)."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"#### 4.3.2.1 The `obj_to_float` function"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The function defined below will convert the given data of type `object` to type `float64`."
]
},
{
"cell_type": "code",
"execution_count": 32,
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"def obj_to_float(data):\n",
" \"\"\"\n",
" Converts the dtype('O') into a dtype('float64')\n",
"\n",
" Parameters\n",
" ----------\n",
" data : Array-like structure\n",
" The data to be converted.\n",
"\n",
" Returns\n",
" -------\n",
" data : np.ndarray\n",
" Returns an np.ndarray\n",
" \"\"\"\n",
"\n",
" try:\n",
" # Converting the data.\n",
" data = data.astype('float64')\n",
"\n",
" # Printing the results to check that the conversion was successful.\n",
" print(f\"New type: {data.dtype}\")\n",
"\n",
" return data\n",
" except Exception as e:\n",
" print(f\"Conversion error: {e}\")\n",
" return None"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"#### 4.3.2.2 Updating the types"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now that I have written the function, it is time to utilise it. I will convert the type of the training and testing labels since they were the ones that had the `object` data type."
]
},
{
"cell_type": "code",
"execution_count": 33,
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"New type: float64\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Converting the training labels to float_64.\n",
"vectorised_train_labels = obj_to_float(vectorised_train_labels)"
]
},
{
"cell_type": "code",
"execution_count": 34,
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"New type: float64\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Converting the testing labels to float_64.\n",
"vectorised_test_labels = obj_to_float(vectorised_test_labels)"
]
},
{
"cell_type": "code",
"execution_count": 35,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"text/plain": [
"(dtype('float64'), dtype('float64'))"
]
},
"execution_count": 35,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Double checking the types\n",
"vectorised_train_labels.dtype, vectorised_test_labels.dtype"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 4.3.3 Splitting the data"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now that the data is of the right type, I will create an instance of the `BuildEvalModel` class, and call the `train_val_split()` method in order to generate training and validation sets. I will then check the shape and type of the data produced in case of error."
]
},
{
"cell_type": "code",
"execution_count": 36,
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"# Creating an instance of the BuildEvalModel class.\n",
"build_eval_model = BuildEvalModel()"
]
},
{
"cell_type": "code",
"execution_count": 37,
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"# Creating training and validation sets (to implement hold-out validation).\n",
"X_train, X_val, y_train, y_val = build_eval_model.train_val_split(vectorised_train_inputs, \n",
" vectorised_train_labels,\n",
" random_seed=1)"
]
},
{
"cell_type": "code",
"execution_count": 38,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"((1440000, 1000),\n",
" dtype('float64'),\n",
" array([[0., 1., 0., ..., 0., 0., 0.],\n",
" [0., 1., 1., ..., 0., 0., 0.],\n",
" [0., 1., 0., ..., 0., 0., 0.],\n",
" ...,\n",
" [0., 1., 1., ..., 0., 0., 0.],\n",
" [0., 1., 0., ..., 0., 0., 0.],\n",
" [0., 1., 1., ..., 0., 0., 0.]]))"
]
},
"execution_count": 38,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Viewing and checking the shape and type of the training data.\n",
"X_train.shape, X_train.dtype, X_train"
]
},
{
"cell_type": "code",
"execution_count": 39,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"text/plain": [
"((1440000,), dtype('float64'), array([1., 0., 0., ..., 0., 1., 1.]))"
]
},
"execution_count": 39,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Viewing and checking the shape and type of the training labels.\n",
"y_train.shape, y_train.dtype, y_train"
]
},
{
"cell_type": "code",
"execution_count": 40,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"((360000, 1000),\n",
" dtype('float64'),\n",
" array([[0., 1., 0., ..., 0., 0., 0.],\n",
" [0., 1., 1., ..., 0., 0., 0.],\n",
" [0., 1., 1., ..., 0., 0., 0.],\n",
" ...,\n",
" [0., 1., 1., ..., 0., 0., 0.],\n",
" [0., 1., 1., ..., 0., 0., 0.],\n",
" [0., 1., 1., ..., 0., 0., 0.]]))"
]
},
"execution_count": 40,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Viewing and checking the shape and type of the validation data.\n",
"X_val.shape, X_val.dtype, X_val"
]
},
{
"cell_type": "code",
"execution_count": 41,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"((360000,), dtype('float64'), array([0., 1., 0., ..., 1., 1., 1.]))"
]
},
"execution_count": 41,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Viewing and checking the shape and type of the validation labels.\n",
"y_val.shape, y_val.dtype, y_val"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "nT70q619NL6k",
"tags": []
},
"source": [
"# 5 Developing a model that does better than the baseline"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Considering that the data is prepared it is time to build the model. The goal is to build the smallest model that does better than statistical power (baseline). In my case statistical power is anything greater than 0.5. This is because I am conducting a binary classification, and therefore have two class labels. In addition, the ratio of positive to negative reviews is equal (see *section 2*)- this makes the common sense baseline 0.5. This is important since the model beating statistical power means that it is doing better than what a human would do.\n",
"\n",
"Table 1 displays the recommended last-layer activation and loss functions based on the problem type."
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "Hx2peubZXXAB"
},
"source": [
"<table>\n",
" <caption><span style=\"font-weight: bold;\">Table 1</span> Choosing the right last-layer activation and loss function for your model [6].</caption>\n",
" <tr style=\"background-color: #ECE5FC;\">\n",
" <th style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\">Problem Type</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\">Last-Layer Activation</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\">Loss Function</th>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\">Binary classification</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\"><code>sigmoid</code></td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\"><code>binary_crossentropy</code></td>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\">Multiclass, single-label classification</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\"><code>softmax</code></td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\"><code>categorical_crossentropy</code></td>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\">Multiclass, multilabel classification</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\"><code>sigmoid</code></td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\"><code>binary_crossentropy</code></td>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\">Regression to arbitrary values</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\">None</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\"><code>mse</code></td>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\">Regression to values between 0 and 1</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\"><code>sigmoid</code></td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: left; padding: 8px;\"><code>mse</code> or <code>binary_crossentropy</code></td>\n",
" </tr>\n",
"</table>"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "bPVMjKkD2gMJ"
},
"source": [
"My problem type is binary classification, so the last-layer activation I will go with is `sigmoid`. The loss function I will use is `binary-crossentropy`. For optimiser configuration, the optimiser I will use is `rmsprop` and its default learning rate."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"## 5.1 The `compile_fit_model` function"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"In order to avoid repetition I will create a function called `compile_fit_model()` that will call the `create_compile_model()` method, the `build_eval_model.fit_model()` method, and return `history.history` (a dictionary that contains a record of training loss and other metrics. This will be especially helpful later on, when it comes to model tuning."
]
},
{
"cell_type": "code",
"execution_count": 42,
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"# I wrote all the code in this cell.\n",
"def compile_fit_model(units: list, activation: list, num_of_layers: int,\n",
" epochs: int = 10, batch_size: int = 512):\n",
" \"\"\"\n",
" A function that compiles and fits a model.\n",
"\n",
" Parameters\n",
" ----------\n",
" units : list of integers (positive)\n",
" List containing the dimensionalities of the output space.\n",
" activation : list of strs\n",
" Specifies the activation functions to be used.\n",
" num_of_layers : int\n",
" The number of layers that the model will have.\n",
" epochs : int, default=10\n",
" The number of epochs to train the model.\n",
" batch_size : int, default=512\n",
" The number of samples per gradient update.\n",
"\n",
" Returns\n",
" -------\n",
" history : dict\n",
" Returns a dict that holds a record of training loss and other \n",
" metrics.\n",
" \"\"\"\n",
" # Creating an instance of the BuildEvalModel class.\n",
" build_eval_model = BuildEvalModel()\n",
"\n",
" # Creating and compiling the model.\n",
" model = build_eval_model.create_compile_model(\n",
" units=units,\n",
" activation=activation,\n",
" input_shape=(1000,),\n",
" num_of_layers=num_of_layers\n",
" )\n",
"\n",
" # Building / fitting the model.\n",
" history = build_eval_model.fit_model(\n",
" model, X_train, y_train, X_val, y_val,\n",
" epochs=epochs,\n",
" batch_size=batch_size\n",
" )\n",
"\n",
" return history.history"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"## 5.2 The `TrainValPlot` class"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"I will also create a class called `TrainValPlot` that will allow me to plot the loss or accuracy of training and validation data, and save time in the long run. The class is contains the following methods:\n",
"1. `display_legend_plot()`\n",
"2. `set_labels(title, xlabel, ylabel)`\n",
"3. `loss_plot(history, colour)`\n",
"4. `accuracy_plot(history, colour)`\n",
"5. `plot_model_loss(hyperparameter_name, hyperparameter, loss, val_loss, colour)`\n",
"6. `plot_model_accuracy(hyperparameter_name, hyperparameter, accuracy, val_accuracy, colour)`\n",
"\n",
"The `display_legend_plot()` method is a helper function that shows the legend and the plot, its purpose is to reduce the amount of repeated code. The `set_labels()` method is also a helper function, it sets the the title and the labels. The `loss_plot()` method plots the training loss against the validation loss. This will enable me to further analyse and visualise the data, it will also help me spot overfitting. The `accuracy_plot()` method plots the training accuracy against the validation accuracy, it will also be useful for spotting the point at which overfitting starts. The `plot_model_loss()` method plots the training loss against the validation loss based on a given hyperparameter. The `plot_model_accuracy()` method plots the training accuracy against the validation accuracy based on a given hyperparameter. The last two methods will be useful when I start tuning the hyperparameters (see *section 7*)."
]
},
{
"cell_type": "code",
"execution_count": 43,
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"#  I wrote all the code in this cell.\n",
"class TrainValPlot:\n",
" \"\"\"\n",
" Class that can plot the loss or accuracy of training and validation data.\n",
"\n",
" Methods\n",
" -------\n",
" display_legend_plot()\n",
" set_labels(title, xlabel, ylabel)\n",
" loss_plot(history, colour)\n",
" accuracy_plot(history, colour)\n",
" plot_model_loss(hyperparameter_name, hyperparameter, loss, val_loss, colour)\n",
" plot_model_accuracy(hyperparameter_name, hyperparameter, accuracy, val_accuracy, colour)\n",
" \"\"\"\n",
"\n",
" def __init__(self):\n",
" pass\n",
"\n",
" def display_legend_plot(self):\n",
" \"\"\"A helper function that shows the legend and the plot.\"\"\"\n",
" plt.legend()\n",
" plt.grid()\n",
" plt.show()\n",
"\n",
" def set_labels(self, title: str, xlabel: str, ylabel: str):\n",
" \"\"\"\n",
" A helper function that sets the title and labels.\n",
"\n",
" Parameters\n",
" ----------\n",
" title : str\n",
" The title of the graph.\n",
" xlabel : str\n",
" The x axis label.\n",
" ylabel : str\n",
" The y axis label.\n",
" \"\"\"\n",
" # Giving a title and labels\n",
" plt.title(title)\n",
" plt.xlabel(xlabel)\n",
" plt.ylabel(ylabel)\n",
"\n",
" def loss_plot(self, history: dict, colour: list = [\"DodgerBlue\", \"Violet\"]):\n",
" \"\"\"\n",
" A function that plots the training loss against the validation loss.\n",
"\n",
" Parameters\n",
" ----------\n",
" history : dict\n",
" The history dictionary that contains information about the loss.\n",
" colour : list, default=[\"DodgerBlue\", \"Violet\"]\n",
" A list of colours (string values).\n",
" \"\"\"\n",
"\n",
" # Accessing the loss values utilising the history dict.\n",
" training_loss = history[\"loss\"]\n",
" val_loss = history[\"val_loss\"]\n",
"\n",
" # Calculating the number of epochs.\n",
" epochs = range(1, len(training_loss) + 1)\n",
"\n",
" #  Plotting the training and validation loss.\n",
" plt.figure(figsize=(10, 6))\n",
" plt.plot(epochs, training_loss, colour[0], label=\"Training Loss\")\n",
" plt.plot(epochs, val_loss, colour[1], label=\"Validation Loss\")\n",
"\n",
" # Setting a title and labels, displaying the plot.\n",
" self.set_labels(\"Training and Validation Loss\", \"Epochs\", \"Loss\")\n",
" self.display_legend_plot()\n",
"\n",
" def accuracy_plot(self, history: dict, colour: list = [\"DodgerBlue\", \"Violet\"]):\n",
" \"\"\"\n",
" A function that plots the training accuracy against the validation\n",
" accuracy.\n",
"\n",
" Parameters\n",
" ----------\n",
" history : dict\n",
" The history dictionary that contains information about the \n",
" accuracy.\n",
" colour : list, default=[\"DodgerBlue\", \"Violet\"]\n",
" A list of colours (string values).\n",
" \"\"\"\n",
"\n",
" # Accessing the accuracy values utilising the history dict.\n",
" training_acc = history[\"accuracy\"]\n",
" val_acc = history[\"val_accuracy\"]\n",
"\n",
" # Calculating the number of epochs.\n",
" epochs = range(1, len(training_acc) + 1)\n",
"\n",
" # Plotting the training and validation accuracy.\n",
" plt.figure(figsize=(10, 6))\n",
" plt.plot(epochs, training_acc, colour[0], label=\"Training Accuracy\")\n",
" plt.plot(epochs, val_acc, colour[1], label=\"Validation Accuracy\")\n",
"\n",
" # Setting a title and labels, displaying the plot.\n",
" self.set_labels(\"Training and Validation Accuracy\",\n",
" \"Epochs\", \"Accuracy\")\n",
" self.display_legend_plot()\n",
"\n",
" def plot_model_loss(self, hyperparameter_name: str, hyperparameter: list,\n",
" loss: list, val_loss: list,\n",
" colour: list = [\"DodgerBlue\", \"Violet\"]):\n",
" \"\"\"\n",
" A function that plots the training loss against the validation\n",
" loss based on a given hyperparameter.\n",
"\n",
" Parameters\n",
" ----------\n",
" hyperparameter_name : str\n",
" The name of the hyperparameter (e.g. the number of units).\n",
" hyperparameter : list\n",
" The list containing the hyperparameter values.\n",
" loss : list\n",
" The loss values.\n",
" val_loss : list\n",
" The validation loss values.\n",
" colour : list, default=[\"DodgerBlue\", \"Violet\"]\n",
" A list of colours (string values).\n",
"\n",
" Notes\n",
" -----\n",
" The inputted lists (hyperparameter, loss, and val_loss) must be the \n",
" same length.\n",
" \"\"\"\n",
" # Checking if the lists are the same length.\n",
" if not (len(hyperparameter) == len(loss) == len(val_loss)):\n",
" raise ValueError(\"Input lists must be the same length.\")\n",
"\n",
" # Plotting the training and validation loss.\n",
" plt.figure(figsize=(10, 6))\n",
" plt.plot(hyperparameter, loss, colour[0], label=\"Training Loss\")\n",
" plt.plot(hyperparameter, val_loss, colour[1], label=\"Validation Loss\")\n",
"\n",
" # Setting a title and labels, displaying the plot.\n",
" self.set_labels(\n",
" f\"{hyperparameter_name}: Training and Validation Loss\", hyperparameter_name, \"Loss\")\n",
" self.display_legend_plot()\n",
"\n",
" def plot_model_accuracy(self, hyperparameter_name: str, hyperparameter: list,\n",
" accuracy: list, val_accuracy: list,\n",
" colour: list = [\"DodgerBlue\", \"Violet\"]):\n",
" \"\"\"\n",
" A function that plots the training accuracy against the validation\n",
" accuracy based on a given hyperparameter.\n",
"\n",
" Parameters\n",
" ----------\n",
" hyperparameter_name : str\n",
" The name of the hyperparameter (e.g. the number of units).\n",
" hyperparameter : list\n",
" The list containing the hyperparameter values.\n",
" accuracy : list\n",
" The accuracy values.\n",
" val_accuracy : list\n",
" The validation accuracy values.\n",
" colour : list, default=[\"DodgerBlue\", \"Violet\"]\n",
" A list of colours (string values).\n",
"\n",
" Notes\n",
" -----\n",
" The inputted lists (hyperparameter, accuracy, and val_accuracy) \n",
" must be the same length.\n",
" \"\"\"\n",
" # Checking if the lists are the same length.\n",
" if not (len(hyperparameter) == len(accuracy) == len(val_accuracy)):\n",
" raise ValueError(\"Input lists must be the same length.\")\n",
"\n",
" # Plotting the training and validation accuracy.\n",
" plt.figure(figsize=(10, 6))\n",
" plt.plot(hyperparameter, accuracy,\n",
" colour[0], label=\"Training Accuracy\")\n",
" plt.plot(hyperparameter, val_accuracy,\n",
" colour[1], label=\"Validation Accuracy\")\n",
"\n",
" # Setting a title and labels, displaying the plot.\n",
" self.set_labels(\n",
" f\"{hyperparameter_name}: Training and Validation Accuracy\", hyperparameter_name, \"Accuracy\")\n",
" self.display_legend_plot()"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"## 5.3 The first model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Since the goal is to beat statistical power and I am aiming to create the simplest model, I will begin with a simple two layer model. Table 2 displays the hyperparameters / parameters I will be using for the model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<table style=\"width: 700px\">\n",
" <caption><span style=\"font-weight: bold;\">Table 2</span> Model 1 hyperparameters / parameters.</caption>\n",
" <tr style=\"background-color: #ECE5FC;\">\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Number of Layers</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Units</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Activation</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Epochs</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Batch Size</th>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">2</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[16, 1]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[\"relu\", \"sigmoid\"]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">5</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">512</td>\n",
" </tr>\n",
"</table>"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 5.3.1 Building the model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"I will begin by using the `compile_fit_model()` function to create, compile, and fit the model."
]
},
{
"cell_type": "code",
"execution_count": 43,
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/5\n",
"2813/2813 [==============================] - 29s 10ms/step - loss: 0.3687 - accuracy: 0.8382 - val_loss: 0.3488 - val_accuracy: 0.8460\n",
"Epoch 2/5\n",
"2813/2813 [==============================] - 23s 8ms/step - loss: 0.3457 - accuracy: 0.8480 - val_loss: 0.3420 - val_accuracy: 0.8496\n",
"Epoch 3/5\n",
"2813/2813 [==============================] - 27s 10ms/step - loss: 0.3412 - accuracy: 0.8502 - val_loss: 0.3394 - val_accuracy: 0.8510\n",
"Epoch 4/5\n",
"2813/2813 [==============================] - 24s 8ms/step - loss: 0.3389 - accuracy: 0.8515 - val_loss: 0.3388 - val_accuracy: 0.8509\n",
"Epoch 5/5\n",
"2813/2813 [==============================] - 26s 9ms/step - loss: 0.3369 - accuracy: 0.8522 - val_loss: 0.3353 - val_accuracy: 0.8524\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Creating, compiling, and fitting the model\n",
"history1a = compile_fit_model(units=[16, 1], \n",
" activation=[\"relu\", \"sigmoid\"], \n",
" num_of_layers=2,\n",
" epochs=5, \n",
" batch_size=512)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Next I will view the keys that are in the dictionary, since these are the keys I used in the `TrainValPlot` class, and I need to ensure that they are what I expect."
]
},
{
"cell_type": "code",
"execution_count": 44,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"text/plain": [
"dict_keys(['loss', 'accuracy', 'val_loss', 'val_accuracy'])"
]
},
"execution_count": 44,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Checking the available keys.\n",
"history1a.keys()"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 5.3.2 Plotting the training and validation loss"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"With the model compiled and fitted, I can now plot the training and validation loss so that I can make comparisons. First I will create an instance of the `TrainValPlot` class (I will be using this instance in the subsequent sections too). Then I will call the `loss_plot()` method."
]
},
{
"cell_type": "code",
"execution_count": 44,
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"# I wrote the code in this cell.\n",
"# Instantiating the class.\n",
"train_val_plot = TrainValPlot()"
]
},
{
"cell_type": "code",
"execution_count": 46,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 1000x600 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Plotting the training and validation loss.\n",
"train_val_plot.loss_plot(history1a)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<caption><span style=\"font-weight: bold;\">Figure 1</span> Training and Validation loss for model 1.</caption>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Figure 1 shows that overfitting doesn't occur, since the validation set does better than the training set. Underfitting doesn't occur either, this is evident since the loss decreases at each epoch meaning that the model is indeed learning the underlying patterns of the data."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 5.3.3 Plotting the training and validation accuracy"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now I will call the `accuracy_plot()` method."
]
},
{
"cell_type": "code",
"execution_count": 47,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 1000x600 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Plotting the training and validation accuracy.\n",
"train_val_plot.accuracy_plot(history1a, [\"SlateBlue\", \"LightGreen\"])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<caption><span style=\"font-weight: bold;\">Figure 2</span> Training and Validation accuracy for model 1.</caption>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Likewise, figure 2 demonstrates that underfitting doesn't occur, and the accuracy generally increases for both training and validation data. However, overfitting may be occurring at the 3rd epoch, since while the training accuracy increases, the validation accuracy drops, though it does rise again at the 5th epoch. The accuracy reaches a high of about 85% for both the training and validation which means that model the achieves statistical power."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"## 5.4 The second model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Although the first model was pretty simple, it still beat statistical power by a large amount. I am going to reduce the model to a 1 layer one (and 1 unit), to see if I am able to achieve statistical power with a smaller model. Table 3 displays the hyperparameters / parameters I will be using for this model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<table style=\"width: 700px\">\n",
" <caption><span style=\"font-weight: bold;\">Table 3</span> Model 2 hyperparameters / parameters.</caption>\n",
" <tr style=\"background-color: #ECE5FC;\">\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Number of Layers</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Units</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Activation</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Epochs</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Batch Size</th>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">1</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[1]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[\"sigmoid\"]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">5</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">512</td>\n",
" </tr>\n",
"</table>"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 5.4.1 Building the model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"I am using the `compile_fit_model()` function to create, compile, and fit the model."
]
},
{
"cell_type": "code",
"execution_count": 48,
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/5\n",
"2813/2813 [==============================] - 32s 11ms/step - loss: 0.4042 - accuracy: 0.8273 - val_loss: 0.3665 - val_accuracy: 0.8409\n",
"Epoch 2/5\n",
"2813/2813 [==============================] - 24s 9ms/step - loss: 0.3657 - accuracy: 0.8415 - val_loss: 0.3644 - val_accuracy: 0.8418\n",
"Epoch 3/5\n",
"2813/2813 [==============================] - 24s 8ms/step - loss: 0.3651 - accuracy: 0.8417 - val_loss: 0.3644 - val_accuracy: 0.8420\n",
"Epoch 4/5\n",
"2813/2813 [==============================] - 23s 8ms/step - loss: 0.3651 - accuracy: 0.8417 - val_loss: 0.3644 - val_accuracy: 0.8420\n",
"Epoch 5/5\n",
"2813/2813 [==============================] - 25s 9ms/step - loss: 0.3652 - accuracy: 0.8418 - val_loss: 0.3644 - val_accuracy: 0.8420\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Creating, compiling, and fitting the model\n",
"history2a = compile_fit_model(units=[1], \n",
" activation=[\"sigmoid\"], \n",
" num_of_layers=1,\n",
" epochs=5, \n",
" batch_size=512)"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 5.4.2 Plotting the training and validation loss"
]
},
{
"cell_type": "code",
"execution_count": 49,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 1000x600 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Plotting the training and validation loss.\n",
"train_val_plot.loss_plot(history2a)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<caption><span style=\"font-weight: bold;\">Figure 3</span> Training and Validation loss for model 2.</caption>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Figure 3 shows that while overfitting doesn't occur, underfitting does. Underfitting starts at the 2nd epoch. The graph demonstrates that from the 2nd epoch, the loss value remains constant, this means that the model isn't learning. This is expected for a simple 1 layer, 1 unit model. The loss values are also higher than model 1."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 5.4.3 Plotting the training and validation accuracy"
]
},
{
"cell_type": "code",
"execution_count": 50,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAA18AAAIjCAYAAAD80aFnAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAACU3klEQVR4nOzdd1wUZ/4H8M/ssiy9SBE7KsZeUQjYsACWcDHJJUYNlkSNuVgiMbEillh+Oc8YE0vuzpKcITFq9MxFRUTBhqJYYo8tGhtFpbct8/uD7IaFpSnLLPB5v16+ZGeemfnOw6L74Zl5RhBFUQQRERERERGZlEzqAoiIiIiIiOoChi8iIiIiIqJqwPBFRERERERUDRi+iIiIiIiIqgHDFxERERERUTVg+CIiIiIiIqoGDF9ERERERETVgOGLiIiIiIioGjB8ERERERERVQOGLyIiMzJ27Fh4eno+07YLFiyAIAhVW5CZ+e233yAIAjZv3lztxxYEAQsWLNC/3rx5MwRBwG+//Vbutp6enhg7dmyV1vM87xUiIpIGwxcRUQUIglChP7GxsVKXWudNnToVgiDgxo0bpbaZO3cuBEHAL7/8Uo2VVd6DBw+wYMECnDt3TupSjLpy5QoEQYCVlRXS0tKkLoeIyOwxfBERVcB//vMfgz+BgYFGl7dt2/a5jvOvf/0L165de6Zt582bh9zc3Oc6fm0watQoAEBkZGSpbb777jt07NgRnTp1eubjhIaGIjc3F82aNXvmfZTnwYMHWLhwodHw9TzvlaqyZcsWeHh4AAC2b98uaS1ERDWBhdQFEBHVBG+99ZbB6xMnTiA6OrrE8uJycnJgY2NT4eMoFIpnqg8ALCwsYGHBf9Z9fX3h5eWF7777DvPnzy+xPj4+Hrdv38by5cuf6zhyuRxyufy59vE8nue9UhVEUURkZCRGjhyJ27dv49tvv8X48eMlrak02dnZsLW1lboMIiKOfBERVZWAgAB06NABiYmJ6NOnD2xsbDBnzhwAwH//+18MHToUDRs2hFKpRMuWLbF48WJoNBqDfRS/j0d3j9OKFSvwz3/+Ey1btoRSqUSPHj1w6tQpg22N3fMlCAImT56MXbt2oUOHDlAqlWjfvj327dtXov7Y2Fh0794dVlZWaNmyJb766qsK30d25MgRvP7662jatCmUSiWaNGmC6dOnlxiJGzt2LOzs7HD//n0MGzYMdnZ2cHNzw4wZM0r0RVpaGsaOHQtHR0c4OTlhzJgxFb60bdSoUbh69SrOnDlTYl1kZCQEQcCIESNQUFCA+fPnw9vbG46OjrC1tUXv3r1x6NChco9h7J4vURTxySefoHHjxrCxsUG/fv1w6dKlEts+efIEM2bMQMeOHWFnZwcHBwcMHjwY58+f17eJjY1Fjx49AADjxo3TX9qqu9/N2D1f2dnZ+PDDD9GkSRMolUq0bt0aK1asgCiKBu0q874ozbFjx/Dbb7/hzTffxJtvvonDhw/j3r17JdpptVp8/vnn6NixI6ysrODm5oZBgwbh9OnTBu22bNkCHx8f2NjYwNnZGX369MH+/fsNai56z51O8fvpdN+XuLg4/O1vf4O7uzsaN24MALhz5w7+9re/oXXr1rC2toaLiwtef/11o/ftpaWlYfr06fD09IRSqUTjxo0xevRopKamIisrC7a2tpg2bVqJ7e7duwe5XI5ly5ZVsCeJqC7hr0iJiKrQ48ePMXjwYLz55pt46623UL9+fQCFHwjt7OwQFhYGOzs7HDx4EPPnz0dGRgb+/ve/l7vfyMhIZGZm4t1334UgCPj000/x6quv4tatW+WOgBw9ehQ//vgj/va3v8He3h6rV6/Ga6+9hrt378LFxQUAcPbsWQwaNAgNGjTAwoULodFosGjRIri5uVXovLdt24acnBy89957cHFxQUJCAr744gvcu3cP27ZtM2ir0WgQHBwMX19frFixAgcOHMA//vEPtGzZEu+99x6AwhDz8ssv4+jRo5g0aRLatm2LnTt3YsyYMRWqZ9SoUVi4cCEiIyPRrVs3g2P/8MMP6N27N5o2bYrU1FT8+9//xogRIzBhwgRkZmZiw4YNCA4ORkJCArp06VKh4+nMnz8fn3zyCYYMGYIhQ4bgzJkzCAoKQkFBgUG7W7duYdeuXXj99dfRvHlzJCUl4auvvkLfvn1x+fJlNGzYEG3btsWiRYswf/58TJw4Eb179wYA+Pv7Gz22KIr4y1/+gkOHDuGdd95Bly5dEBUVhY8++gj379/HZ599ZtC+Iu+Lsnz77bdo2bIlevTogQ4dOsDGxgbfffcdPvroI4N277zzDjZv3ozBgwdj/PjxUKvVOHLkCE6cOIHu3bsDABYuXIgFCxbA398fixYtgqWlJU6ePImDBw8iKCiowv1f1N/+9je4ublh/vz5yM7OBgCcOnUKx48fx5tvvonGjRvjt99+w7p16xAQEIDLly/rR6mzsrLQu3dvXLlyBW+//Ta6deuG1NRU7N69G/fu3UOXLl3wyiuvYOvWrVi5cqXBCOh3330HURT1l78SERkQiYio0t5//32x+D+hffv2FQGI69evL9E+JyenxLJ3331XtLGxEfPy8vTLxowZIzZr1kz/+vbt2yIA0cXFRXzy5Il++X//+18RgPjTTz/pl0VERJSoCYBoaWkp3rhxQ7/s/PnzIgDxiy++0C8LCQkRbWxsxPv37+uXXb9+XbSwsCixT2OMnd+yZctEQRDEO3fuGJwfAHHRokUGbbt27Sp6e3vrX+/atUsEIH766af6ZWq1Wuzdu7cIQNy0aVO5NfXo0UNs3LixqNFo9Mv27dsnAhC/+uor/T7z8/MNtnv69KlYv3598e233zZYDkCMiIjQv960aZMIQLx9+7YoiqKYnJwsWlpaikOHDhW1Wq2+3Zw5c0QA4pgxY/TL8vLyDOoSxcLvtVKpNOibU6dOlXq+xd8ruj775JNPDNr99a9/FQVBMHgPVPR9UZqCggLRxcVFnDt3rn7ZyJEjxc6dOxu0O3jwoAhAnDp1aol96Pro+vXrokwmE1955ZUSfVK0H4v3v06zZs0M+lb3fenVq5eoVqsN2hp7n8bHx4sAxG+++Ua/bP78+SIA8ccffyy17qioKBGAuHfvXoP1nTp1Evv27VtiOyIiURRFXnZIRFSFlEolxo0bV2K5tbW1/uvMzEykpqaid+/eyMnJwdWrV8vd7/Dhw+Hs7Kx/rRsFuXXrVrnbDhw4EC1bttS/7tSpExwcHPTbajQaHDhwAMOGDUPDhg317by8vDB48OBy9w8Ynl92djZSU1Ph7+8PURRx9uzZEu0nTZpk8Lp3794G57Jnzx5YWFjoR8KAwnuspkyZUqF6gML79O7du4fDhw/rl0VGRsLS0hKvv/66fp+WlpYACi+Pe/LkCdRqNbp37270ksWyHDhwAAUFBZgyZYrBpZoffPBBibZKpRIyWeF/wRqNBo8fP4adnR1at25d6ePq7NmzB3K5HFOnTjVY/uGHH0IURezdu9dgeXnvi7Ls3bsXjx8/xogRI/TLRowYgfPnzxtcZrljxw4IgoCIiIgS+9D10a5du6DVajF//nx9nxRv8ywmTJhQ4p68ou9TlUqFx48fw8vLC05OTgb9vmPHDnTu3BmvvPJKqXUPHDgQDRs2xLfffqtfd/HiRfzyyy/l3gtKRHUXwxcRURVq1KiR/sN8UZcuXcIrr7wCR0dHODg4wM3NTf8BLT09vdz9Nm3a1OC1Log9ffq00tvqttdtm5ycjNzcXHh5eZVoZ2yZMXfv3sXYsWNRr149/X1cffv2BVDy/HT3/ZRWD1B4b06DBg1gZ2dn0K5169YVqgcA3nzzTcjlcv2sh3l5edi5cycGDx5sEGS//vprdOrUCVZWVnBxcYGbmxt+/vnnCn1firpz5w4AoFWrVgbL3dzcDI4HFAa9zz77DK1atYJSqYSrqyvc3Nzwyy+/VPq4RY/fsGFD2NvbGyzXzcCpq0+nvPdFWbZs2YLmzZtDqVTixo0buHHjBlq2bAkbGxuDMHLz5k00bNgQ9erVK3VfN2/ehEwmQ7t27co9bmU0b968xLLc3FzMnz9ff0+crt/T0tIM+v3mzZvo0KFDmfuXyWQYNWoUdu3ahZycHACFl2JaWVnpwz0RUXEMX0REVajob9Z10tLS0LdvX5w/fx6LFi3CTz/9hOjoaPzf//0fgMIP4uUpbVY9sdhEClW9bUVoNBoEBgbi559/xsyZM7Fr1y5ER0frJ4Yofn7VNUOgu7s7AgMDsWPHDqhUKvz000/IzMw0uBdny5YtGDt2LFq2bIkNGzZg3759iI6ORv/+/Sv0fXlWS5cuRVhYGPr06YMtW7YgKioK0dHRaN++vUmPW9Szvi8yMjLw008/4fbt22jVqpX+T7t27ZCTk4PIyMgqe29VRPGJWnSM/SxOmTIFS5YswRtvvIEffvgB+/fvR3R0NFxcXJ6p30ePHo2srCzs2rVLP/vjSy+9BEdHx0rvi4jqBk64QURkYrGxsXj8+DF+/PFH9OnTR7/89u3bElb1J3d3d1hZWRl9KHFZDyrWuXDhAn799Vd8/fXXGD16tH55dHT0M9fUrFkzxMTEICsry2D0q7LPtRo1ahT27duHvXv3IjIyEg4ODggJCdGv3759O1q0aIEff/zR4BI3Y5fJVaRmALh+/TpatGihX56SklJiNGn79u3o168fNmzYYLA8LS0Nrq6u+teVueyuWbNmOHDgADIzMw1Gv3SXtVbV88h+/PFH5OXlYd26dQa1AoXfn3nz5uHYsWPo1asXWrZsiaioKDx58qTU0a+WLVtCq9Xi8uXLZU5w4uzsXGK2y4KCAjx8+LDCtW/fvh1jxozBP/7xD/2yvLy8Evtt2bIlLl68WO7+OnTogK5du+Lbb79F48aNcffuXXzxxRcVroeI6h6OfBERmZhuhKHoaEBBQQHWrl0rVUkG5HI5Bg4ciF27duHBgwf65Tdu3Chxn1Bp2wOG5yeKIj7//PNnrmnIkCFQq9VYt26dfplGo6n0B9thw4bBxsYGa9euxd69e/Hqq6/CysqqzNpPnjyJ+Pj4Stc8cOBAKBQKfPHFFwb7W7VqVYm2crm8xOjQtm3bcP/+fYNlumdTVWSK/SFDhkCj0eDLL780WP7ZZ59BEIQK379Xni1btqBFixaYNGkS/vrXvxr8mTFjBuzs7PSXHr722msQRRELFy4ssR/d+Q8bNgwymQyLFi0qMfpUtI9atmxpcP8eAPzzn/8sdeTLGGP9/sUXX5TYx2uvvYbz589j586dpdatExoaiv3792PVqlVwcXGpsn4motqJI19ERCbm7+8PZ2dnjBkzBlOnToUgCPjPf/5TrZdmlWfBggXYv38/evbsiffee0//Ib5Dhw44d+5cmdu2adMGLVu2xIwZM3D//n04ODhgx44dFbp3qDQhISHo2bMnZs2ahd9++w3t2rXDjz/+WOn7oezs7DBs2DD9fV/Fp/9+6aWX8OOPP+KVV17B0KFDcfv2baxfvx7t2rVDVlZWpY6le17ZsmXL8NJLL2HIkCE4e/Ys9u7dW2KE6KWXXsKiRYswbtw4+Pv748KFC/j2228NRsyAwsDh5OSE9evXw97eHra2tvD19TV6P1NISAj69euHuXPn4rfffkPnzp2xf/9+/Pe//8UHH3xgMLnGs3rw4AEOHTpUYlIPHaVSieDgYGzbtg2rV69Gv379EBoaitWrV+P69esYNGgQtFotjhw5gn79+mHy5Mnw8vLC3LlzsXjxYvTu3RuvvvoqlEolTp06hYYNG+qflzV+/HhMmjQJr732GgIDA3H+/HlERUWV6NuyvPTSS/jPf/4DR0dHtGvXDvHx8Thw4ECJqfU/+ugjbN++Ha+//jrefvtteHt748mTJ9i9ezfWr1+Pzp0769uOHDkSH3/8MXbu3In33ntP8odfE5F548gXEZGJubi44H//+x8aNGiAefPmYcWKFQgMDMSnn34qdWl63t7e2Lt3L5ydnREeHo4NGzZg0aJFGDBggMFIkTEKhQI//fQTunTpgmXLlmHhwoVo1aoVvvnmm2euRyaTYffu3Rg1ahS2bNmCuXPnolGjRvj6668rvS9d4GrQoAH69+9vsG7s2LFYunQpzp8/j6lTpyIqKgpbtmzRP3+qsj755BMsXLgQZ8+exUcffYSbN29i//79+hEsnTlz5uDDDz9EVFQUpk2bhjNnzuDnn39GkyZNDNopFAp8/fXXkMvlmDRpEkaMGIG4uDijx9b12QcffID//e9/+OCDD3D58mX8/e9/x8qVK5/pfIr7/vvvodVqDS7dLC4kJASPHz/Wj5pu2rQJf//733H79m189NFHWLp0KXJzcw2eV7Zo0SJs3LgRubm5mDt3LubPn487d+5gwIAB+jYTJkzAzJkzcfjwYXz44Ye4ffs2oqOjS/RtWT7//HOMHj0a3377LT788EM8fPgQBw4cKDGxi52dHY4cOYL33nsPe/bswdSpU7F27Vq0bt1a/8Bmnfr16+ufRRYaGlrhWoiobhJEc/rVKxERmZVhw4bh0qVLuH79utSlEJmtV155BRcuXKjQPZJEVLdx5IuIiAAUTsNd1PXr17Fnzx4EBARIUxBRDfDw4UP8/PPPHPUiogrhyBcREQEovCxv7NixaNGiBe7cuYN169YhPz8fZ8+eLfHsKqK67vbt2zh27Bj+/e9/49SpU7h58yY8PDykLouIzBwn3CAiIgDAoEGD8N133+HRo0dQKpXw8/PD0qVLGbyIjIiLi8O4cePQtGlTfP311wxeRFQhHPkiIiIiIiKqBrzni4iIiIiIqBowfBEREREREVUD3vP1jLRaLR48eAB7e3sIgiB1OUREREREJBFRFJGZmYmGDRtCJit9fIvh6xk9ePCgxMMwiYiIiIio7vr9999LPIy9KIavZ2Rvbw+gsIMdHBwkrUWlUmH//v0ICgqCQqGQtJbaiP1rWuxf02L/mhb717TYv6bHPjYt9q9pmVP/ZmRkoEmTJvqMUBqGr2eku9TQwcHBLMKXjY0NHBwcJH/j1UbsX9Ni/5oW+9e02L+mxf41PfaxabF/Tcsc+7e825E44QYREREREVE1YPgiIiIiIiKqBgxfRERERERE1YDhi4iIiIiIqBowfBEREREREVUDhi8iIiIiIqJqwPBFRERERERUDRi+iIiIiIiIqgHDFxERERERUTVg+CIiIiIiIqoGDF9ERERERETVgOGLiIiIiIioGjB8ERERERERVQOGLyIiIiIiomrA8EVERERERFQNzCJ8rVmzBp6enrCysoKvry8SEhLKbL9q1Sq0bt0a1tbWaNKkCaZPn468vDyjbZcvXw5BEPDBBx/olz158gRTpkzR76Np06aYOnUq0tPTq/K0iIiIiIiI9CykLmDr1q0ICwvD+vXr4evri1WrViE4OBjXrl2Du7t7ifaRkZGYNWsWNm7cCH9/f/z6668YO3YsBEHAypUrDdqeOnUKX331FTp16mSw/MGDB3jw4AFWrFiBdu3a4c6dO5g0aRIePHiA7du3m/R8iYiIqOqJoggRYuHXxf42tky/Tix9XfHtDZaLqNA2Rv8Wy9hvWduWc8xS6xVL369ao0a2SzZuqW9BLsohQAAA/d86glByeXlfG10vGG9T4njl7K+8miq1v1JqeqZzLHJYAQLUohpamRYqUQVBrNw56s6PahfJw9fKlSsxYcIEjBs3DgCwfv16/Pzzz9i4cSNmzZpVov3x48fRs2dPjBw5EgDg6emJESNG4OTJkwbtsrKyMGrUKPzrX//CJ598YrCuQ4cO2LFjh/51y5YtsWTJErz11ltQq9WwsJC8W4iIqBqIoggttFCLaqihhlpUQwUV1KIaGlGj/1q3rvjXBdoCpLZKxaG8QxAKCj8olfYBuqwPwBX50F3qfsvatoqOWaltK3DMygQLTU8Nvsr6qtw66Dm0B/bl7ZO6itqrF/Cv7H891y7MJWAChqGwvFrK3Z9Qfk1l7U8UReS0yEFNImnKKCgoQGJiImbPnq1fJpPJMHDgQMTHxxvdxt/fH1u2bEFCQgJ8fHxw69Yt7NmzB6GhoQbt3n//fQwdOhQDBw4sEb6MSU9Ph4ODQ6nBKz8/H/n5+frXGRkZAACVSgWVSlXu/k1Jd3yp66it2L+mxf41rZrYv6IoQg01NNBAJaqggcYw9ODPr42uK9JGI2qMriu67+f+AN8AuKK+UjUnTyXJAQ00kpZQ0Q+SZY3mCBDK/dBa3vZGlwlGjlOJ7UVRRFpaGpycnAxGbYwF7NK+Lv4zZCxMV8k+jX1tJOyXta/n+VpKz11TVZ6GeXSJnrWNtVn8H1fRGiQNX6mpqdBoNKhfv77B8vr16+Pq1atGtxk5ciRSU1PRq1evwv+g1WpMmjQJc+bM0bf5/vvvcebMGZw6darCdSxevBgTJ04stc2yZcuwcOHCEsv3798PGxubCh3H1KKjo6UuoVZj/5oW+9e0nrd/RYgQ5SJEmQitTKv/WpQVW15sWYXaF133x2tJiICgFSBohMK/tQJkGpn+66LrZFrZn69FweDDSPFLiyq6Dvjjg3Hx0xefY31ltzVSk+610fVlrauK9SjWZ0a2F1BF6419L+oIa1hLXUKNYTT4GAutxd8+JX70xZJtTNGuovWZW7sy2hb/HshVckRflP4zRE5OxUbgatz1dbGxsVi6dCnWrl0LX19f3LhxA9OmTcPixYsRHh6O33//HdOmTUN0dDSsrKzK3V9GRgaGDh2Kdu3aYcGCBaW2mz17NsLCwgy2a9KkCYKCguDg4FAVp/bMVCoVoqOjERgYCIVCIWkttRH717TYv89GFMXCUZ8yRnfUUCNfnY8Lly/ghbYvQJSLJUaOjG5vZJ0WWknOUwYZLGABC8FC/7ccciiggFyQl1hngcL1xZcV397YOhlklb7Hgu9f02L/mh772LTYv6ZlTv2ruyquPJKGL1dXV8jlciQlJRksT0pKgoeHh9FtwsPDERoaivHjxwMAOnbsiOzsbEycOBFz585FYmIikpOT0a1bN/02Go0Ghw8fxpdffon8/HzI5XIAQGZmJgYNGgR7e3vs3LmzzG+aUqmEUqkssVyhUEj+zdYxp1pqI/avadWG/tVdLqcSVfrgovu6xP1CpdxDVOZ9R8W2r7AXgHhNPKrqyq0SwaW0r4ssU0DxZ3ASFBXaxgIWkAlmMSlvuWrD+9ecsX9Nj31sWuzfZ6fVitBoRGg1hX8Xfg1oNCLy80Xk5cjNon8renxJw5elpSW8vb0RExODYcOGAQC0Wi1iYmIwefJko9vk5ORAJjP8z1gXpkRRxIABA3DhwgWD9ePGjUObNm0wc+ZMfduMjAwEBwdDqVRi9+7dFRolI6Kax2BU5zmCTlnbF73/SAoChD9HgooEHQvBAjJRhsdJj9HIoxEsZZZlBp2iXysEhX4ESSEoDEaNOAMXEZH5EEXDQPJnQBGNvta1K75eoxELg45a9zWgUYv68GPwtabkMbVGjqv/WrcvjQiN1nBfxkJV0eWikSs9i3Jt4IxX/1o9fV0VJL/sMCwsDGPGjEH37t3h4+ODVatWITs7Wz/74ejRo9GoUSMsW7YMABASEoKVK1eia9eu+ssOw8PDERISArlcDnt7e3To0MHgGLa2tnBxcdEvz8jIQFBQEHJycrBlyxZkZGTohwrd3Nz0AY2ITEsUReSJeSiwLcAjzSMAqFDQUYkqfahSiSp98DE2yiTVzdKlXvpWRtCpzIhQ0a/lQun/ZqlUKuy5vAdBnkGS/1aQiEgKovjHh3qDYCGWEixKfvg3HixQargpun+VWoNbN52QlXwfgFBKe5R9rDKCi0ZdfjipjQQBkFsIkMsFyGQ1qwMkD1/Dhw9HSkoK5s+fj0ePHqFLly7Yt2+ffhKOu3fvGox0zZs3D4IgYN68ebh//z7c3NwQEhKCJUuWVPiYZ86c0U9N7+XlZbDu9u3b8PT0fP4TIyIAfwasNG0a0jRp+r/TtelI06YhX8wHvIEfc380eS3PeulbRUaEim/P0SEiqilEsTBwQNSWM2oCI6MjxUdEUGKUxHiwMBylMVhXzuiLRlu8PpQILIZBSOoetsfvN55U6xEFAYXBRF4YUAq/RpGvdcFFKAwxsj+WWQiQy4xtK5S5P4N9Ffla9se+Cr8WDNob7rfkMY2/BmSywn0Bf/yCcc+eau3b5yWIYl3My88vIyMDjo6O+inqpaR74w0ZMoS/2TYB9m/5RFFErphrGKx0QUubhgKxoMztZQUy2Cnt/gwxZQQduSA3DD3l3Hekuxyvrl4ux/dv2QpnzRWhKtCiIF+Lgj/+VhWI+q//XFb4ddG2eblq3LlzDw0bNvrz/SUa/PXna1EsZXnJmgxfl9Ou2PH07UrUIRquL75/IwtLbVvsJEo/pmGtJZaX0k5HqxWRkZEBe3sHg5/fip+78XbF+77CfVTR8zZSQ0X7qPRjFmtX2rmX067kuRcGobroz1BQ8Q/+hsFCMAwWRYKLbhkEEbdv3cALbVrBUmFRImwYDSIGIQZGwpHh8UsLVXWBOf0fV9FsIPnIFxHVDKIoIkfMMQxWRYJWAcoOWHaCHZzkTnCSOcFJ7gRHmSOc5E6w0dgg+nC0WfzDSeajtFBUGH7EckPRn211XxfbV5Ftn/9XkLZ4eCetCs6ajLNEZlqe1EXUSfoP9aWFjVI++FckuJQ56lLqsXT7Q5lB6c8A8uelafrlfxxTEFAtv5ArDAdnMGhIff4fRwAYvoioCF3AKjpqla5J1/9dXsCyl9nDSfZnsCoatCwE4//cqLTSPxiRKu5ZQlHxr6svFFWOTA5YWspgqZQV/m0pg+KPrxV/vNav++NvmVzEjZvX0K5dW8jlMugeTFP8M53+taD7SyhlebHXglDK8uL7F4q9Lq2d4ReC0XXFayn7nIRixVW6XRnnrlarcepUAnx8fGBhYVHK/ivZR8WOV9r3pvh+SvveFD9PY8sqeu6lfn/KOadSv88VaKdSq3HoUAyCggbCysrSILjUxasFiEyN4YuojhFFEdlitj5UFb0XK12TDhXKDkO6gKULVk4yJzjKHcsMWGR6pYWinJwCPE5S4tK5DGg0sjoTihSWAiyVxtcXXab/2lIGuUXlP2iqVCqo95xGQLArf6ttAiqVCrd/z0ebDvbsXxNRqQQorbSwtbOAQsEJx4hMjZ+UiGohXcAqenlgmvbPSwTLekaUAOHPESy5o0HQcpA5MGBVkiiKUKtEqFQlR4qK31tUsVBUbISpQqHIHYlxd6r83ORyQR9yiociS6UMCoWRUKSUwdJSMHkoIiIiMkf8FEVUQ4miiCwxy+gEF+ma9HIDloPMocTlgU4yJ9jL7OtEwHrWUFQyCD1PKDINuVw36iNAoZAhvyAL9VwcoVTKnykU6UeYGIqIiIieS+3/hEVUg4miiEwx889LBIsFrbIe6qsLWMUvD3SSFY5glfVsKCmVH4oKL4mrTCgyCFNmEoqMjRQZu4xOUfQyumcIRX/OBOXNy7aIiIgkxvBFJDGtqEWWNqvEBBe6oFVWwJJBVuYIlrkGrKKOHHiMIz83wPG9l/X3GUkVioqGnIqGosKRJKGUkSSOFBEREdGfGL6IqoEuYD3VPjUYuUrXpFc4YBUNVroZBGtKwCpNTrYaP217hIJ8C8BIH1QkFBW/jI6hiIiIiMwVwxdRFdGKWmRqM41OcJGuTYcW2lK3lUGmH71ylDmWGMGSCbJqPJPqczz2MQrytbB1UGHyzHawsbFkKCIiIqJai+GLqBK0ohYZ2owSDxpO06YhQ5tRZsCSQw5HmeOfMwj+EbScZc6wk9nV2oBVGq1WxOHoFABAsxcy0aCRFe9JIiIiolqN4YuoGI2oMRjBeqJ+gqQOSfg2+1tkipkVCljFJ7hwkjnVyYBVll8S0/E4pQC2dnI0aJojdTlEREREJsfwRXWSRtQgQ5tRYoIL3QiWiGIzPtQDcsVcAIUBq+jIVdGgZS/YQxB4qVxFHNqXDADwD6gHweI3aYshIiIiqgYMX1RrFQ1YRSe4KDVgFWEBC/2olT3s8ful39GrUy+4KF1gJ9gxYD2n33/LwY2rWZDJgZ4DXHA8XuqKiIiIiEyP4YtqNLWoLgxYxSa4SNOmIVObWW7AMjbBhZPcCbaCrT5gqVQq7Hm0B426NYJCxnuSqoJu1KubjzOcnNmnREREVDcwfJHZKx6wigat8gKWAgqjE1w4yh0NAhZVn4x0FRLjnwIAAga5S1wNERERUfVh+CKzoBbVJWYQTNem60ewyqKAwugEF05yJ9gINgxYZubIgVSo1SKae9miuZctVCqV1CURERERVQuGL6o2KlFl8NyrokErS8wqc1tLWBqd4MJJxoBVk6hUWhyJKZxePmCQm8TVEBEREVUvhi+qUkUDlsEIViUClrGZBK0FawasWuDMiafITFfDyVmBrj2cpS6HiIiIqFoxfFGlqUTVn+Gq2FTt2WJ2mdtaCpYGlwUWDVoMWLWbKIr6iTb6BLpBbsHvNREREdUtDF9kVIFYUOL5V7qgVV7AUgpKw3BV5BJBK8GKAauOunktG7//lguFQkDP/q5Sl0NERERU7Ri+6rB8Mb9EwNJdIpgj5pS5rZVgZXh5YJGwZSWzqqYzoJpEN+rl06se7Oz5Tw8RERHVPfwEVMvli/lGJ7hI06YhV8wtc1trwdrgvquiQYsBiyrjcUo+zp9OA8Dp5YmIiKjuYviqBfLFfOTb5eO66joy1ZkGQasiAavoqFXRqdqVMmU1nQHVdnHRKRBFoE0HezRsbC11OURERESSYPiq4bSiFpuyN0HbTYuH+Q+NtrERbEqMYOmCllJgwCLTysvT4PihxwCAfhz1IiIiojqM4auGkwkyOAqOyMjLgLu1O5wtnOEoc4SzvPBvBiySWsKRJ8jN0cDdQ4l2nR2kLoeIiIhIMgxftcAbNm8gKi4KQ4YMgUKhkLocIj2tVkRsVOFEG32D3CCTcaZLIiIiqrtkUhdAz08uyKUugcioK79kIOlhPqysZXixj4vU5RARERFJiuGLiEzmUFQKAMA/wBVW1vwlAREREdVtDF9EZBKP7ufiyi8ZEITCSw6JiIiI6jqGLyIyCd2oVydvR7i6c9IXIiIiIoYvIqpy2VlqnDzC6eWJiIiIimL4IqIqd/xQKlQFIho1tYZXGzupyyEiIiIyCwxfRFSlNBoRcdGFlxz2G+QOQeD08kREREQAwxcRVbHzp9Pw9LEKdg4W6O7nLHU5RERERGaD4YuIqtShvYUPVe49wBUKS/4TQ0RERKTDT0ZEVGXu3MrGrevZkMsF9B7I6eWJiIiIimL4IqIqc2hf4b1e3n7OcHRSSFwNERERkXlh+CKiKpH+VIUzJ54C4PTyRERERMYwfBFRlTgSkwKNRkSLF2zRtLmN1OUQERERmR2GLyJ6bqoCLY7EpALgqBcRERFRaRi+iOi5nY5/iqwMNZxdLNG5u5PU5RARERGZJYYvInouoiji0L7C6eX7BrlBLudDlYmIiIiMYfgioudy/UoW7t/NhaVShp79XKQuh4iIiMhsMXwR0XPRjXr59q4HG1sLiashIiIiMl8MX0T0zFKT83HhTDoAICCYE20QERERlYXhi4ieWWxUCkQRaNfJAR4NraQuh4iIiMismUX4WrNmDTw9PWFlZQVfX18kJCSU2X7VqlVo3bo1rK2t0aRJE0yfPh15eXlG2y5fvhyCIOCDDz4wWJ6Xl4f3338fLi4usLOzw2uvvYakpKSqOiWiWi83R4P42MLp5QMGuUlcDREREZH5kzx8bd26FWFhYYiIiMCZM2fQuXNnBAcHIzk52Wj7yMhIzJo1CxEREbhy5Qo2bNiArVu3Ys6cOSXanjp1Cl999RU6depUYt306dPx008/Ydu2bYiLi8ODBw/w6quvVvn5EdVWJw4/Rl6eFvUbKtG2o4PU5RARERGZPcnD18qVKzFhwgSMGzcO7dq1w/r162FjY4ONGzcabX/8+HH07NkTI0eOhKenJ4KCgjBixIgSo2VZWVkYNWoU/vWvf8HZ2dlgXXp6OjZs2ICVK1eif//+8Pb2xqZNm3D8+HGcOHHCZOdKVFtotSLi9qcAKLzXSybj9PJERERE5ZF0arKCggIkJiZi9uzZ+mUymQwDBw5EfHy80W38/f2xZcsWJCQkwMfHB7du3cKePXsQGhpq0O7999/H0KFDMXDgQHzyyScG6xITE6FSqTBw4ED9sjZt2qBp06aIj4/Hiy++WOK4+fn5yM/P17/OyMgAAKhUKqhUqsqffBXSHV/qOmor9m9Jl85lICUpH9Y2MnR70eG5+ob9a1rsX9Ni/5oW+9f02Memxf41LXPq34rWIGn4Sk1NhUajQf369Q2W169fH1evXjW6zciRI5GamopevXpBFEWo1WpMmjTJ4LLD77//HmfOnMGpU6eM7uPRo0ewtLSEk5NTieM+evTI6DbLli3DwoULSyzfv38/bGxsyjrNahMdHS11CbUa+/dPp2PdAFihfpM0xMTsq5J9sn9Ni/1rWuxf02L/mh772LTYv6ZlDv2bk5NToXY17qE8sbGxWLp0KdauXQtfX1/cuHED06ZNw+LFixEeHo7ff/8d06ZNQ3R0NKysqm72tdmzZyMsLEz/OiMjA02aNEFQUBAcHKS930WlUiE6OhqBgYFQKBSS1lIbsX8NPbyXh/0/XIcgAKMn9EA9V8vn2h/717TYv6bF/jUt9q/psY9Ni/1rWubUv7qr4sojafhydXWFXC4vMctgUlISPDw8jG4THh6O0NBQjB8/HgDQsWNHZGdnY+LEiZg7dy4SExORnJyMbt266bfRaDQ4fPgwvvzyS+Tn58PDwwMFBQVIS0szGP0q67hKpRJKpbLEcoVCIfk3W8ecaqmN2L+Fjhx4AADo0sMJ9RvYVtl+2b+mxf41LfavabF/TY99bFrsX9Myh/6t6PElnXDD0tIS3t7eiImJ0S/TarWIiYmBn5+f0W1ycnIgkxmWLZfLAQCiKGLAgAG4cOECzp07p//TvXt3jBo1CufOnYNcLoe3tzcUCoXBca9du4a7d++WelwiArIy1Th17AkAIGAQH6pMREREVBmSX3YYFhaGMWPGoHv37vDx8cGqVauQnZ2NcePGAQBGjx6NRo0aYdmyZQCAkJAQrFy5El27dtVfdhgeHo6QkBDI5XLY29ujQ4cOBsewtbWFi4uLfrmjoyPeeecdhIWFoV69enBwcMCUKVPg5+dndLINIip09GAqVCoRTZvboOULVTfqRURERFQXSB6+hg8fjpSUFMyfPx+PHj1Cly5dsG/fPv0kHHfv3jUY6Zo3bx4EQcC8efNw//59uLm5ISQkBEuWLKnUcT/77DPIZDK89tpryM/PR3BwMNauXVul50ZUm2jUIg5H66aXd4MgcHp5IiIiosqQPHwBwOTJkzF58mSj62JjYw1eW1hYICIiAhERERXef/F9AICVlRXWrFmDNWvWVKZUojrrbMJTpD9VwcHJAt1edC5/AyIiIiIyIPlDlomoZji0LxkA0HuAGxQK/tNBREREVFn8BEVE5bp9PRu/3cyBhYWA3gNcpS6HiIiIqEZi+CKich2KKhz16u7vDHtHTpVLRERE9CwYvoioTE8fF+DsyacAgIBgTi9PRERE9KwYvoioTIcPpECrBVq1tUMTTxupyyEiIiKqsRi+iKhUBflaHDuYCgDox4cqExERET0Xhi8iKtWpY0+QnaWBi5slOnZzlLocIiIiohqN4YuIjBJFUT/RRt8gN8hkfKgyERER0fNg+CIio65dysTDe3lQWsngH8Dp5YmIiIieF8MXERmle6jyi31cYG0jl7gaIiIiopqP4YuISkh+lIeLZzMAAAHBbhJXQ0RERFQ7MHwRUQmxUSkAgA5dHODuYSVxNURERES1A8MXERnIzdHgxOHHAIB+gzm9PBEREVFVYfgiIgPxcanIz9OiQWMrtG5vL3U5RERERLUGwxcR6Wm1ov6Sw37B7hAETi9PREREVFUYvohI78KZdDxOKYCtnRw9etaTuhwiIiKiWoXhi4j0dNPL9+zvCksl/3kgIiIiqkr8dEVEAIDff8vB9StZkMmAPgM5vTwRERFRVWP4IiIAQGxU4ahXV19nOLtYSlwNERERUe3D8EVEyExX4fTxpwAKJ9ogIiIioqrH8EVEOHowFWq1iGYtbdC8la3U5RARERHVSgxfRHWcWq3F4QOF08v3H8RRLyIiIiJTYfgiquPOnHiKjDQ1HJ0V6OrjLHU5RERERLUWwxdRHSaKIg7tKxz16hPoBrkFH6pMREREZCoMX0R12K1fs3H3dg4UCgG9+rtKXQ4RERFRrcbwRVSH6R6q3KNnPdjZW0hcDREREVHtxvBFVEc9TsnHuVNpAIAATi9PREREZHIMX0R11OHoFIgi0Lq9PRo1tZa6HCIiIqJaj+GLqA7Kz9Pg2KHHAIB+g9wkroaIiIiobmD4IqqDTh59gtwcDVzrK9G+i6PU5RARERHVCQxfRHWMVisiNqpwoo2AIDfIZJxenoiIiKg6MHwR1TFXLmQg6UE+rKxl8OvrInU5RERERHUGwxdRHRP7x0OV/fq6wspaLnE1RERERHUHwxdRHfLoQR4u/5IBQQACgjnRBhEREVF1YvgiqkN093p17OYIV3elxNUQERER1S0MX0R1RE62GiePPAEA9ONDlYmIiIiqHcMXUR1x7NBjFORr0aipNVq1s5O6HCIiIqI6h+GLqA7QaETE7S+caCMg2A2CwOnliYiIiKobwxdRHXD+dBqePi6Anb0FevjXk7ocIiIiojqJ4YuoDtBNtNFrgCsUlvyxJyIiIpICP4UR1XJ3b+fg5rVsyOUCeg90lbocIiIiojqL4Yuolju0r3DUq9uLTnBytpS4GiIiIqK6i+GLqBZLf6pCYvxTAEC/QZxenoiIiEhKDF9EtdiRmBRoNCJatLJFsxa2UpdDREREVKcxfBHVUqoCLY7EpAIA+g3mqBcRERGR1Bi+iGqp0/FPkZWhhrOLAp27O0ldDhEREVGdJ3n4WrNmDTw9PWFlZQVfX18kJCSU2X7VqlVo3bo1rK2t0aRJE0yfPh15eXn69evWrUOnTp3g4OAABwcH+Pn5Ye/evQb7ePToEUJDQ+Hh4QFbW1t069YNO3bsMMn5EUlBFEX99PJ9At0gl/OhykRERERSkzR8bd26FWFhYYiIiMCZM2fQuXNnBAcHIzk52Wj7yMhIzJo1CxEREbhy5Qo2bNiArVu3Ys6cOfo2jRs3xvLly5GYmIjTp0+jf//+ePnll3Hp0iV9m9GjR+PatWvYvXs3Lly4gFdffRVvvPEGzp49a/JzJqoON65m4d6dXCgsBfTsx+nliYiIiMyBpOFr5cqVmDBhAsaNG4d27dph/fr1sLGxwcaNG422P378OHr27ImRI0fC09MTQUFBGDFihMFoWUhICIYMGYJWrVrhhRdewJIlS2BnZ4cTJ04Y7GfKlCnw8fFBixYtMG/ePDg5OSExMdHk50xUHXTTy/v2coGtnYXE1RARERERAEj2qaygoACJiYmYPXu2fplMJsPAgQMRHx9vdBt/f39s2bIFCQkJ8PHxwa1bt7Bnzx6EhoYaba/RaLBt2zZkZ2fDz8/PYD9bt27F0KFD4eTkhB9++AF5eXkICAgotd78/Hzk5+frX2dkZAAAVCoVVCpVZU69yumOL3UdtVVN69/HKQX4JTEdANBrgLPZ113T+remYf+aFvvXtNi/psc+Ni32r2mZU/9WtAZBFEXRxLUY9eDBAzRq1AjHjx83CEYff/wx4uLicPLkSaPbrV69GjNmzIAoilCr1Zg0aRLWrVtn0ObChQvw8/NDXl4e7OzsEBkZiSFDhujXp6WlYfjw4di/fz8sLCxgY2ODbdu2ISgoqNR6FyxYgIULF5ZYHhkZCRsbm8qePpHJXDvnhDu/2sOlfi68+6ZKXQ4RERFRrZeTk4ORI0ciPT0dDg4OpbarUdcjxcbGYunSpVi7di18fX1x48YNTJs2DYsXL0Z4eLi+XevWrXHu3Dmkp6dj+/btGDNmDOLi4tCuXTsAQHh4ONLS0nDgwAG4urpi165deOONN3DkyBF07NjR6LFnz56NsLAw/euMjAw0adIEQUFBZXZwdVCpVIiOjkZgYCAUCoWktdRGNal/83I1OPzTVQBavDaqLdp1spe6pHLVpP6tidi/psX+NS32r+mxj02L/Wta5tS/uqviyiNZ+HJ1dYVcLkdSUpLB8qSkJHh4eBjdJjw8HKGhoRg/fjwAoGPHjsjOzsbEiRMxd+5cyGSFt7BZWlrCy8sLAODt7Y1Tp07h888/x1dffYWbN2/iyy+/xMWLF9G+fXsAQOfOnXHkyBGsWbMG69evN3pspVIJpVJZYrlCoZD8m61jTrXURjWhf48deoq8XC3cGyjRsaszZLKaM8thTejfmoz9a1rsX9Ni/5oe+9i02L+mZQ79W9HjSzbhhqWlJby9vRETE6NfptVqERMTY3AZYlE5OTn6gKUjl8sBFE6tXRqtVqu/XysnJwcAjO5Hq9VW/kSIzIRWKyIuKgUA0C/YvUYFLyIiIqK6QNLLDsPCwjBmzBh0794dPj4+WLVqFbKzszFu3DgAhVPCN2rUCMuWLQNQOJPhypUr0bVrV/1lh+Hh4QgJCdGHsNmzZ2Pw4MFo2rQpMjMzERkZidjYWERFRQEA2rRpAy8vL7z77rtYsWIFXFxcsGvXLkRHR+N///ufNB1BVAUun89A8qN8WNvI4dO7ntTlEBEREVExkoav4cOHIyUlBfPnz8ejR4/QpUsX7Nu3D/Xr1wcA3L1712CEat68eRAEAfPmzcP9+/fh5uaGkJAQLFmyRN8mOTkZo0ePxsOHD+Ho6IhOnTohKioKgYGBAAqHBPfs2YNZs2YhJCQEWVlZ8PLywtdff20wKQdRTaObXt6/nwusrOQSV0NERERExUk+4cbkyZMxefJko+tiY2MNXltYWCAiIgIRERGl7m/Dhg3lHrNVq1bYsWNHpeokMmcP7uXi6sVMCALQN9BN6nKIiIiIyAhJH7JMRFUj9o9Rr87dneDiVnJiGCIiIiKSHsMXUQ2XlalGwtEnAIB+g9wlroaIiIiISsPwRVTDHT+UCpVKRBNPa7RsbSt1OURERERUCoYvohpMoxYRt/+P6eUHuUMQOL08ERERkbli+CKqwc6deoq0pyrYO1qg24vOUpdDRERERGVg+CKqwQ7tKxz16j3ADQoFf5yJiIiIzBk/rRHVULdvZOP2jWxYWAjoPdBV6nKIiIiIqBwMX0Q1lG56eW8/Zzg4KiSuhoiIiIjKw/BFVAOlPSnAmYSnADi9PBEREVFNwfBFVAMdPpACrQbwamOHJp42UpdDRERERBXA8EVUwxQUaHE0JhUA0G+Qm8TVEBEREVFFMXwR1TCnjj1BdpYGLm6W6OTtJHU5RERERFRBDF9ENYgoivqJNvoEukEm40OViYiIiGoKhi+iGuTapUw8uJcHS6UM/gEuUpdDRERERJXA8EVUg8RGFT5U+cU+LrCxtZC4GiIiIiKqDIYvohoi+VEeLp5NBwAEBHGiDSIiIqKahuGLqIaI258CUQTad3FA/YZWUpdDRERERJXE8EVUA+TmaHAi7jEAoF8wH6pMREREVBMxfBHVAPFxj5GXp4VHQyu06WgvdTlERERE9AwYvojMnFYrIm5/4fTyAYPcIAicXp6IiIioJmL4IjJzF86kIzW5ADa2cvj24vTyRERERDUVwxeRmYuNKhz16tnfFZZK/sgSERER1VT8JEdkxu7fzcGvl7MgkwF9Azm9PBEREVFNxvBFZMYO7St8qHKXHk5wdrGUuBoiIiIieh4MX0RmKjNDhVPHnwAA+g3m9PJERERENR3DF5GZOhqTCrVKRLMWNmjuZSt1OURERET0nBi+iMyQWq3F4QOFlxz2G+TO6eWJiIiIagGGLyIzdOZkGjLS1HB0UqCrr5PU5RARERFRFWD4IjIzoiji0L7C6eV7D3SFhQV/TImIiIhqA36qIzIzt69n4+6tHFgoBPQa4Cp1OURERERURRi+iMyMbtSrh3892DsoJK6GiIiIiKoKwxeRGXn6uADnTqUBAPoN4kOViYiIiGoThi8iMxIXnQKtFnihnR0aNbWRuhwiIiIiqkIMX0RmIj9Pg2MHUwEUTi9PRERERLULwxeRmUg49gQ52Rq4uluiQ1dHqcshIiIioirG8EVkBkRRROy+wocqBwS7QybjQ5WJiIiIahuGLyIzcPVCJh49yIOVlQwv9nGRuhwiIiIiMgGGLyIzcPCP6eX9AlxgbSOXuBoiIiIiMgWGLyKJJT3Iw+XzGRAEoG8QJ9ogIiIiqq0YvogkFhtVOOrVoasj3OorJa6GiIiIiEyF4YtIQjnZapw48gQAp5cnIiIiqu0YvogkdDz2MQrytWjY2AovtLOTuhwiIiIiMiGGLyKJaDQi4vYXTi/fb7A7BIHTyxMRERHVZgxfRBK5cCYNT1ILYGsnR3f/elKXQ0REREQmxvBFJJGDewtHvXoPcIOlJX8UiYiIiGo7fuIjksDd2zm4eS0LMjnQO9BV6nKIiIiIqBpIHr7WrFkDT09PWFlZwdfXFwkJCWW2X7VqFVq3bg1ra2s0adIE06dPR15enn79unXr0KlTJzg4OMDBwQF+fn7Yu3dvif3Ex8ejf//+sLW1hYODA/r06YPc3NwqPz8iY3TTy3fzdYaTs6XE1RARERFRdZA0fG3duhVhYWGIiIjAmTNn0LlzZwQHByM5Odlo+8jISMyaNQsRERG4cuUKNmzYgK1bt2LOnDn6No0bN8by5cuRmJiI06dPo3///nj55Zdx6dIlfZv4+HgMGjQIQUFBSEhIwKlTpzB58mTIZJJnUaoD0tNUSIx/CoDTyxMRERHVJRZSHnzlypWYMGECxo0bBwBYv349fv75Z2zcuBGzZs0q0f748ePo2bMnRo4cCQDw9PTEiBEjcPLkSX2bkJAQg22WLFmCdevW4cSJE2jfvj0AYPr06Zg6darBMVq3bl3l50dkzNGYFKjVIpq3soVnS1upyyEiIiKiaiJZ+CooKEBiYiJmz56tXyaTyTBw4EDEx8cb3cbf3x9btmxBQkICfHx8cOvWLezZswehoaFG22s0Gmzbtg3Z2dnw8/MDACQnJ+PkyZMYNWoU/P39cfPmTbRp0wZLlixBr169Sq03Pz8f+fn5+tcZGRkAAJVKBZVKVenzr0q640tdR21Vlf2rVmlx+EAqAKD3wHr8noHvX1Nj/5oW+9e02L+mxz42LfavaZlT/1a0BkEURdHEtRj14MEDNGrUCMePH9cHIwD4+OOPERcXZzCaVdTq1asxY8YMiKIItVqNSZMmYd26dQZtLly4AD8/P+Tl5cHOzg6RkZEYMmQIAODEiRPw8/NDvXr1sGLFCnTp0gXffPMN1q5di4sXL6JVq1ZGj7tgwQIsXLiwxPLIyEjY2Ng8azdQHXP/NxtcSnCB0lqN3kMfgle6EhEREdV8OTk5GDlyJNLT0+Hg4FBqO0kvO6ys2NhYLF26FGvXroWvry9u3LiBadOmYfHixQgPD9e3a926Nc6dO4f09HRs374dY8aMQVxcHNq1awetVgsAePfdd/WXO3bt2hUxMTHYuHEjli1bZvTYs2fPRlhYmP51RkYGmjRpgqCgoDI7uDqoVCpER0cjMDAQCoVC0lpqo6rqX1EU8Y8FNwDkIXBoIwx8qWvVFVmD8f1rWuxf02L/mhb71/TYx6bF/jUtc+pf3VVx5ZEsfLm6ukIulyMpKclgeVJSEjw8PIxuEx4ejtDQUIwfPx4A0LFjR2RnZ2PixImYO3eufsIMS0tLeHl5AQC8vb1x6tQpfP755/jqq6/QoEEDAEC7du0M9t22bVvcvXu31HqVSiWUSmWJ5QqFQvJvto451VIbPW//Xr+SiXt38qCwFNB7YH0oFDXqdx8mx/evabF/TYv9a1rsX9NjH5sW+9e0zKF/K3p8yS56srS0hLe3N2JiYvTLtFotYmJiDC5DLConJ6fEjIRyuRxA4ahCabRarf5+LU9PTzRs2BDXrl0zaPPrr7+iWbNmz3QuRBURG1X4UGWfXi6ws2fwIiIiIqprJP0EGBYWhjFjxqB79+7w8fHBqlWrkJ2drb8ccPTo0WjUqJH+UsCQkBCsXLkSXbt21V92GB4ejpCQEH0Imz17NgYPHoymTZsiMzMTkZGRiI2NRVRUFABAEAR89NFHiIiIQOfOndGlSxd8/fXXuHr1KrZv3y5NR1Ct9zglH+dPpwEAAoLdpC2GiIiIiCQhafgaPnw4UlJSMH/+fDx69AhdunTBvn37UL9+fQDA3bt3DUa65s2bB0EQMG/ePNy/fx9ubm4ICQnBkiVL9G2Sk5MxevRoPHz4EI6OjujUqROioqIQGBiob/PBBx8gLy8P06dPx5MnT9C5c2dER0ejZcuW1XfyVKfE7U+BKAJtOtijYWNrqcshIiIiIglIfu3T5MmTMXnyZKPrYmNjDV5bWFggIiICERERpe5vw4YNFTrurFmzjD5LjKiq5eVpcDz2MQA+VJmIiIioLuNE10QmdvLwE+TmaODuoUS7ztLOjElERERE0mH4IjIhrVZE7P5kAEDfYDfIZILEFRERERGRVCodvjw9PbFo0aIyp2UnokKXf8lA8sN8WNvI8WIfF6nLISIiIiIJVTp8ffDBB/jxxx/RokULBAYG4vvvv9dP405EhmL3FY56+fV1gZWVXOJqiIiIiEhKzxS+zp07h4SEBLRt2xZTpkxBgwYNMHnyZJw5c8YUNRLVSA/v5eLKhUwIAtA3iNPLExEREdV1z3zPV7du3bB69Wo8ePAAERER+Pe//40ePXqgS5cu2LhxY5kPPSaqC3QPVe7k7QhXd6XE1RARERGR1J55qnmVSoWdO3di06ZNiI6Oxosvvoh33nkH9+7dw5w5c3DgwAFERkZWZa1ENUZ2lhonj3J6eSIiIiL6U6XD15kzZ7Bp0yZ89913kMlkGD16ND777DO0adNG3+aVV15Bjx49qrRQoprk2KFUqApENG5mDa82dlKXQ0RERERmoNLhq0ePHggMDMS6deswbNgwKBSKEm2aN2+ON998s0oKJKppNGoRcfsLLznsN8gdgsDp5YmIiIjoGcLXrVu30KxZszLb2NraYtOmTc9cFFFNdu50GtKeqGDvYAFvP2epyyEiIiIiM1HpCTeSk5Nx8uTJEstPnjyJ06dPV0lRRDXZoT+ml+81wBUKBZ9jTkRERESFKv3J8P3338fvv/9eYvn9+/fx/vvvV0lRRDXVnZvZuH09G3K5gN4DOb08EREREf2p0uHr8uXL6NatW4nlXbt2xeXLl6ukKKKa6lBU4aiXt58zHJ1K3g9JRERERHVXpcOXUqlEUlJSieUPHz6EhcUzz1xPVOOlPS1A4omnADi9PBERERGVVOnwFRQUhNmzZyM9PV2/LC0tDXPmzEFgYGCVFkdUkxw5kAqtBmjZ2hZNm9tIXQ4RERERmZlKD1WtWLECffr0QbNmzdC1a1cAwLlz51C/fn385z//qfICiWoCVYEWR2NSAXDUi4iIiIiMq3T4atSoEX755Rd8++23OH/+PKytrTFu3DiMGDHC6DO/iOqCU8efICtTDWcXS3TydpK6HCIiIiIyQ890k5atrS0mTpxY1bUQ1UiiKOqnlw8IdoNczocqExEREVFJzzxDxuXLl3H37l0UFBQYLP/LX/7y3EUR1STXL2fhwe95sFTK4B/gInU5RERERGSmKh2+bt26hVdeeQUXLlyAIAgQRREAIAiFv+3XaDRVWyGRmTv4x6jXi73rwcaWM34SERERkXGVnu1w2rRpaN68OZKTk2FjY4NLly7h8OHD6N69O2JjY01QIpH5SknKx8WzhTN/BgRzog0iIiIiKl2lf00fHx+PgwcPwtXVFTKZDDKZDL169cKyZcswdepUnD171hR1EpmluP3JEEWgXWcH1G9oJXU5RERERGTGKj3ypdFoYG9vDwBwdXXFgwcPAADNmjXDtWvXqrY6IjOWm6NBfOxjAJxenoiIiIjKV+mRrw4dOuD8+fNo3rw5fH198emnn8LS0hL//Oc/0aJFC1PUSGSWThx+jLw8Leo3VKJtR3upyyEiIiIiM1fp8DVv3jxkZ2cDABYtWoSXXnoJvXv3houLC7Zu3VrlBRKZI61WROz+FABAv2B3/YQzRERERESlqXT4Cg4O1n/t5eWFq1ev4smTJ3B2duYHUKozLp1LR2pSPqxt5PDpVU/qcoiIiIioBqjUPV8qlQoWFha4ePGiwfJ69eoxeFGdcnBf4ahXr/6uUFrJJa6GiIiIiGqCSoUvhUKBpk2b8lleVKfdv5uLXy9lQiYD+gS6SV0OEREREdUQlZ7tcO7cuZgzZw6ePHliinqIzF5sVOFDlTv3cEI9V0uJqyEiIiKimqLS93x9+eWXuHHjBho2bIhmzZrB1tbWYP2ZM2eqrDgic5OZoULCscJfPPTjQ5WJiIiIqBIqHb6GDRtmgjKIaoZjB1OhVolo2sIGLV6wLX8DIiIiIqI/VDp8RUREmKIOIrOnVmtx+EAqAKBfsBsnmSEiIiKiSqn0PV9EddX5UxlIf6qCg5MFur3oLHU5RERERFTDVHrkSyaTlfkbf86ESLWRKAJx+wtHvfoMdIOFBX9vQURERESVU+nwtXPnToPXKpUKZ8+exddff42FCxdWWWFE5iT9sSXu3s6FhUJAr/6uUpdDRERERDVQpcPXyy+/XGLZX//6V7Rv3x5bt27FO++8UyWFEZmTu9ftAQDd/evB3lEhcTVEREREVBNV2bVTL774ImJiYqpqd0Rm4+njAiTdswZQONEGEREREdGzqJLwlZubi9WrV6NRo0ZVsTsis3L04BOIogCvNrZo3MxG6nKIiIiIqIaq9GWHzs7OBhNuiKKIzMxM2NjYYMuWLVVaHJHUCvK1iI8tfKhyn0AXiashIiIiopqs0uHrs88+MwhfMpkMbm5u8PX1hbMzp9+m2iXh2BPkZGtgbatGh64OUpdDRERERDVYpcPX2LFjTVAGkfkRRRGH9iUDAJp6ZUIm40OViYiIiOjZVfqer02bNmHbtm0llm/btg1ff/11lRRFZA6uXszEo/t5UFrJ0LB5ttTlEBEREVENV+nwtWzZMri6lnzOkbu7O5YuXVolRRGZg9g/Rr18ejlDYSlKXA0RERER1XSVDl93795F8+bNSyxv1qwZ7t69WyVFEUkt6WEeLp7LgCAAfQZyog0iIiIien6VDl/u7u745ZdfSiw/f/48XFz4IZVqh7j9KQCA9l0c4OahlLgaIiIiIqoNKh2+RowYgalTp+LQoUPQaDTQaDQ4ePAgpk2bhjfffNMUNRJVq5xsNeLjHgMA+g1yl7gaIiIiIqotKh2+Fi9eDF9fXwwYMADW1tawtrZGUFAQ+vfv/8z3fK1Zswaenp6wsrKCr68vEhISymy/atUqtG7dGtbW1mjSpAmmT5+OvLw8/fp169ahU6dOcHBwgIODA/z8/LB3716j+xJFEYMHD4YgCNi1a9cz1U+1S3zcYxTka9GwsRVat7eXuhwiIiIiqiUqPdW8paUltm7dik8++QTnzp2DtbU1OnbsiGbNmj1TAVu3bkVYWBjWr18PX19frFq1CsHBwbh27Rrc3UuOOkRGRmLWrFnYuHEj/P398euvv2Ls2LEQBAErV64EADRu3BjLly9Hq1atIIoivv76a7z88ss4e/Ys2rdvb7C/VatWGTy3jOo2rVbUX3IYMMid7w0iIiIiqjKVDl86rVq1QqtWrZ67gJUrV2LChAkYN24cAGD9+vX4+eefsXHjRsyaNatE++PHj6Nnz54YOXIkAMDT0xMjRozAyZMn9W1CQkIMtlmyZAnWrVuHEydOGISvc+fO4R//+AdOnz6NBg0aPPe5UM33S2I6HqcUwNZOjh4960ldDhERERHVIpUOX6+99hp8fHwwc+ZMg+WffvopTp06ZfQZYKUpKChAYmIiZs+erV8mk8kwcOBAxMfHG93G398fW7ZsQUJCAnx8fHDr1i3s2bMHoaGhRttrNBps27YN2dnZ8PPz0y/PycnByJEjsWbNGnh4eJRba35+PvLz8/WvMzIyAAAqlQoqlapC52squuNLXUdtcHBvEgDAL6AeBEEDlUrD/jUx9q9psX9Ni/1rWuxf02Mfmxb717TMqX8rWoMgimKlHmDk5uaGgwcPomPHjgbLL1y4gIEDByIpKanC+3rw4AEaNWqE48ePGwSjjz/+GHFxcQajWUWtXr0aM2bMgCiKUKvVmDRpEtatW1eiHj8/P+Tl5cHOzg6RkZEYMmSIfv27774LjUaDf//73wAAQRCwc+dODBs2zOgxFyxYgIULF5ZYHhkZCRsbmwqfM5mvjKcKnIj2gCCI6D30IaxsNFKXREREREQ1gG5gJz09HQ4ODqW2q/TIV1ZWFiwtLUssVygU+tEgU4qNjcXSpUuxdu1a+Pr64saNG5g2bRoWL16M8PBwfbvWrVvj3LlzSE9Px/bt2zFmzBjExcWhXbt22L17Nw4ePIizZ89W+LizZ89GWFiY/nVGRgaaNGmCoKCgMju4OqhUKkRHRyMwMBAKhULSWmqyyH/fA/AUXXyc8OpfO+mXs39Ni/1rWuxf02L/mhb71/TYx6bF/jUtc+rfiuagSoevjh07YuvWrZg/f77B8u+//x7t2rWr1L5cXV0hl8tLjJYlJSWVeilgeHg4QkNDMX78eH092dnZmDhxIubOnQuZrHACR0tLS3h5eQEAvL29cerUKXz++ef46quvcPDgQdy8eRNOTk4G+37ttdfQu3dvxMbGljiuUqmEUlnyeU8KhULyb7aOOdVS02Skq3DmRBoAYMAQD6P9yP41LfavabF/TYv9a1rsX9NjH5sW+9e0zKF/K3r8Soev8PBwvPrqq7h58yb69+8PAIiJiUFkZCS2b99eqX1ZWlrC29sbMTEx+sv9tFotYmJiMHnyZKPb5OTk6AOWjlwuB1A4bXxptFqt/p6tWbNm6cObTseOHfHZZ5+VmKyD6oajMalQq0V4trRBcy9bqcshIiIiolqo0uErJCQEu3btwtKlS7F9+3ZYW1ujc+fOOHjwIOrVq/zscGFhYRgzZgy6d+8OHx8frFq1CtnZ2frZD0ePHo1GjRph2bJl+uOvXLkSXbt21V92GB4ejpCQEH0Imz17NgYPHoymTZsiMzMTkZGRiI2NRVRUFADAw8PD6Mha06ZN0bx580qfA9VsKpUWhw8UTi/fbzAfqkxEREREpvFMU80PHToUQ4cOBVB4feN3332HGTNmIDExERpN5SYpGD58OFJSUjB//nw8evQIXbp0wb59+1C/fn0AwN27dw1GuubNmwdBEDBv3jzcv38fbm5uCAkJwZIlS/RtkpOTMXr0aDx8+BCOjo7o1KkToqKiEBgY+CynS7XcmRNPkZmuhpOzAl17OEtdDhERERHVUs/8nK/Dhw9jw4YN2LFjBxo2bIhXX30Va9aseaZ9TZ48udTLDIvff2VhYYGIiAhERESUur8NGzZUuoZKTvpItYQoiji0LxkA0CfQDXILPlSZiIiIiEyjUuHr0aNH2Lx5MzZs2ICMjAy88cYbyM/Px65duyo92QaRObj5azZ+/y0XCoWAnv1dpS6HiIiIiGoxWflNCoWEhKB169b45ZdfsGrVKjx48ABffPGFKWsjMrlDewtHvXx61YOd/TMPBBMRERERlavCnzb37t2LqVOn4r333kOrVq1MWRNRtXicko/zp9MAAAHBnGiDiIiIiEyrwiNfR48eRWZmJry9veHr64svv/wSqamppqyNyKTiolMgikCbDvZo2MRa6nKIiIiIqJarcPh68cUX8a9//QsPHz7Eu+++i++//x4NGzaEVqtFdHQ0MjMzTVknUZXKy9Pg+KHHADjqRURERETVo8LhS8fW1hZvv/02jh49igsXLuDDDz/E8uXL4e7ujr/85S+mqJGoyiUceYLcHA3c6ivRvouD1OUQERERUR1Q6fBVVOvWrfHpp5/i3r17+O6776qqJiKT0mpFxEYVTrQREOwGmYzTyxMRERGR6T1X+NKRy+UYNmwYdu/eXRW7IzKpKxcykPQwH1bWMrzYx0XqcoiIiIiojqiS8EVUkxzalwIA8A9whZW1XOJqiIiIiKiuYPiiOuXR/Vxc+SUDggD0DXKTuhwiIiIiqkMYvqhOORRVOOrVydsRru5KiashIiIiorqE4YvqjOwsNRKOPgHA6eWJiIiIqPoxfFGdcfxQKgrytWjU1Bqt2tpJXQ4RERER1TEMX1QnaDQi4qILLznsN8gdgsDp5YmIiIioejF8UZ1w/nQanj5Wwc7BAt39nKUuh4iIiIjqIIYvqhMO7St8qHLvAa5QWPJtT0RERETVj59Cqda7cysbt37NhlwuoPcATi9PRERERNJg+KJaT/dQ5W4vOsPRWSFxNURERERUVzF8Ua2W/lSFMyeeAgD6DeKoFxERERFJh+GLarUjMSnQaES0eMEWzVrYSl0OEREREdVhDF9Ua6kKtDgSkwqgcHp5IiIiIiIpMXxRrXU6/imyMtRwdlGgc3cnqcshIiIiojqO4YtqJVEU9dPL9w10g1zOhyoTERERkbQYvqhWun4lC/fv5sJSKYN/P1epyyEiIiIiYvii2ik2qnDUy6dXPdjaWUhcDRERERERwxfVQqnJ+fglMR0A0C+Y08sTERERkXlg+KJaJ25/CkQRaNvJAR6NrKUuh4iIiIgIAMMX1TJ5uRocj9VNL89RLyIiIiIyHwxfVKucOPwYebla1G+gRNuODlKXQ0RERESkx/BFtYZWKyI2KgUAEDDIHTIZp5cnIiIiIvPB8EW1xqVzGUhJyoe1jRy+vepJXQ4RERERkQGGL6o1dA9V9u/nAqWVXOJqiIiIiIgMMXxRrfDg91xcu5QJQQD6BnKiDSIiIiIyPwxfVCvoRr06d3eCi5tS4mqIiIiIiEpi+KIaLytTjVPHngAA+g12l7gaIiIiIiLjGL6oxjt2MBUqlYgmntZo+YKt1OUQERERERnF8EU1mkYt4nB04fTy/Qa5QxA4vTwRERERmSeGL6rRzp56irSnKtg7WqDbi85Sl0NEREREVCqGL6rRDu0tnGijz0A3KBR8OxMRERGR+eKnVaqxbl/Pxm83c2BhIaDXAFepyyEiIiIiKhPDF9VYh6IKR726+zvDwVEhcTVERERERGVj+KIa6enjApxNeAoACAjm9PJEREREZP4YvqhGOnwgBVoN0KqtHZp42khdDhERERFRuRi+qMYpyNfi2MFUABz1IiIiIqKag+GLapxTx54gO0sDFzdLdPJ2lLocIiIiIqIKYfiiGkUURf1EG32D3CCT8aHKRERERFQzMHxRjXLtUiYe3suDpVIGv74uUpdDRERERFRhZhG+1qxZA09PT1hZWcHX1xcJCQlltl+1ahVat24Na2trNGnSBNOnT0deXp5+/bp169CpUyc4ODjAwcEBfn5+2Lt3r379kydPMGXKFP0+mjZtiqlTpyI9Pd1k50hV49C+wlEvv74usLG1kLgaIiIiIqKKkzx8bd26FWFhYYiIiMCZM2fQuXNnBAcHIzk52Wj7yMhIzJo1CxEREbhy5Qo2bNiArVu3Ys6cOfo2jRs3xvLly5GYmIjTp0+jf//+ePnll3Hp0iUAwIMHD/DgwQOsWLECFy9exObNm7Fv3z6888471XLO9GySH+Xh0rkMAIWXHBIRERER1SSSDx2sXLkSEyZMwLhx4wAA69evx88//4yNGzdi1qxZJdofP34cPXv2xMiRIwEAnp6eGDFiBE6ePKlvExISYrDNkiVLsG7dOpw4cQLt27dHhw4dsGPHDv36li1bYsmSJXjrrbegVqthYSF5t5ARsVEpEEWgQxcH1G9gJXU5RERERESVImnKKCgoQGJiImbPnq1fJpPJMHDgQMTHxxvdxt/fH1u2bEFCQgJ8fHxw69Yt7NmzB6GhoUbbazQabNu2DdnZ2fDz8yu1lvT0dDg4OJQavPLz85Gfn69/nZFROAKjUqmgUqnKPVdT0h1f6jpMKTdHgxOHHwMAeg+sV63nWhf6V0rsX9Ni/5oW+9e02L+mxz42LfavaZlT/1a0BkEURdHEtZTqwYMHaNSoEY4fP24QjD7++GPExcUZjGYVtXr1asyYMQOiKEKtVmPSpElYt26dQZsLFy7Az88PeXl5sLOzQ2RkJIYMGWJ0f6mpqfD29sZbb72FJUuWGG2zYMECLFy4sMTyyMhI2NjwIb+mdueaHa6dd4atgwr+wY8gcJJDIiIiIjITOTk5GDlypH5ApzQ17vq62NhYLF26FGvXroWvry9u3LiBadOmYfHixQgPD9e3a926Nc6dO4f09HRs374dY8aMQVxcHNq1a2ewv4yMDAwdOhTt2rXDggULSj3u7NmzERYWZrBdkyZNEBQUVGYHVweVSoXo6GgEBgZCoVBIWospaLUilhy6BkCFoa96wj+gW7Uev7b3r9TYv6bF/jUt9q9psX9Nj31sWuxf0zKn/tVdFVceScOXq6sr5HI5kpKSDJYnJSXBw8PD6Dbh4eEIDQ3F+PHjAQAdO3ZEdnY2Jk6ciLlz50ImK5xDxNLSEl5eXgAAb29vnDp1Cp9//jm++uor/b4yMzMxaNAg2NvbY+fOnWV+05RKJZRKZYnlCoVC8m+2jjnVUpXOn07D4xQVbO3k8OvjBoVCmnliamv/mgv2r2mxf02L/Wta7F/TYx+bFvvXtMyhfyt6fElnO7S0tIS3tzdiYmL0y7RaLWJiYkq9PysnJ0cfsHTkcjmAwgfwlkar1Za4ZysoKAiWlpbYvXs3rKw4gYO50k0v37OfKyyVkk/QSURERET0TCS/7DAsLAxjxoxB9+7d4ePjg1WrViE7O1s/++Ho0aPRqFEjLFu2DEDhTIYrV65E165d9ZcdhoeHIyQkRB/CZs+ejcGDB6Np06bIzMxEZGQkYmNjERUVBeDP4JWTk4MtW7YgIyNDP1To5uam3w9J7/ffcnD9ShZkMqBPIKeXJyIiIqKaS/LwNXz4cKSkpGD+/Pl49OgRunTpgn379qF+/foAgLt37xqMdM2bNw+CIGDevHm4f/8+3NzcEBISYjBRRnJyMkaPHo2HDx/C0dERnTp1QlRUFAIDAwEAZ86c0U/mobs0Uef27dvw9PQ08VlTRcXuTwEAdPVxhrOLpcTVEBERERE9O8nDFwBMnjwZkydPNrouNjbW4LWFhQUiIiIQERFR6v42bNhQ5vECAgLKvESRzENmugqnjz0BAAQM4qgXEREREdVsvIGGzNbRg6lQq0U0a2mD5l62UpdDRERERPRcGL7ILKnVWhw+UHjJYb9gdwh8sBcRERER1XAMX2SWzpx4iow0NRydFejq6yR1OUREREREz43hi8yOKIo4FFU46tVnoCssLPg2JSIiIqKaj59qyezc+jUbd2/lQKEQ0LO/q9TlEBERERFVCYYvMjuHogofqtyjZz3YO/Bp8ERERERUOzB8kVl5klqA86fSAAABwe7SFkNEREREVIUYvsisHI5OgVYLvNDeHo2aWktdDhERERFRlWH4IrORn6fB0YOpAID+fKgyEREREdUyDF9kNhKOPkFujgau9ZVo38VR6nKIiIiIiKoUwxeZBa1W1E+0ERDkBpmMD1UmIiIiotqF4YvMwtWLmUh6kA8rKxle7OMidTlERERERFWO4YvMwqG9haNefgEusLaRS1wNEREREVHVY/giyT16kIfLv2RAEIC+QZxenoiIiIhqJ4YvklzsH/d6dejqCLf6SomrISIiIiIyDYYvklROthonjzwBAPQfxFEvIiIiIqq9GL5IUsdjH6MgX4tGTa3Rqp2d1OUQEREREZkMwxdJRqMRERuVAgAICHaDIHB6eSIiIiKqvRi+SDK/JKbh6eMC2NlboId/PanLISIiIiIyKYYvksyhfYUTbfQa4AqFJd+KRERERFS78RMvSeLu7RzcvJYNmRzoPdBV6nKIiIiIiEyO4YskoRv18n7RGU7OlhJXQ0RERERkegxfVO3S01RIjH8KAOgXzOnliYiIiKhuYPiianfkQAo0GhEtWtmiWUtbqcshIiIiIqoWDF9UrVQFWhyNSQUABPChykRERERUhzB8UbU6Hf8UmRlqONVToEt3J6nLISIiIiKqNgxfVG1EUURsVOFEG32D3CC34EOViYiIiKjuYPiianPjahbu3cmFwlJAz36cXp6IiIiI6haGL6o2uunlfXu5wNbOQuJqiIiIiIiqF8MXVYvU5Hz8kpgOAAgIdpO4GiIiIiKi6sfwRdUibn8KRBFo29EeDRpbS10OEREREVG1Y/gik8vL1SA+7jEATi9PRERERHUXwxeZ3Ikjj5Gbo4F7AyXadXKQuhwiIiIiIkkwfJFJabUi4qJSAAABQe6QyTi9PBERERHVTQxfZFKXz2cg+VE+rG3k8O1TT+pyiIiIiIgkw/BFJqWbXt4/wAVWVnKJqyEiIiIikg7DF5nMg3u5uHoxE4IA9A3i9PJEREREVLcxfJHJxP5xr1fn7k5wcVNKXA0RERERkbQYvsgksjLVSDjyx/TyfKgyERERERHDF5nG8UOpUKlENPG0hlcbO6nLISIiIiKSHMMXVTmNWkTc/j+mlw92hyBwenkiIiIiIoYvqnLnTj1F2lMV7B0s4O3nLHU5RERERERmgeGLqtyhPyba6D3QFQoF32JERERERADDF1Wx325m4/b1bFhYCOg1gBNtEBERERHpMHxRldI9VNnbzxmOTgqJqyEiIiIiMh8MX1Rl0p4U4MzJpwAKJ9ogIiIiIqI/mUX4WrNmDTw9PWFlZQVfX18kJCSU2X7VqlVo3bo1rK2t0aRJE0yfPh15eXn69evWrUOnTp3g4OAABwcH+Pn5Ye/evQb7yMvLw/vvvw8XFxfY2dnhtddeQ1JSkknOr644ciAVWg3QsrUdmja3kbocIiIiIiKzInn42rp1K8LCwhAREYEzZ86gc+fOCA4ORnJystH2kZGRmDVrFiIiInDlyhVs2LABW7duxZw5c/RtGjdujOXLlyMxMRGnT59G//798fLLL+PSpUv6NtOnT8dPP/2Ebdu2IS4uDg8ePMCrr75q8vOtrQoKtDgSUzjRRv/BvNeLiIiIiKg4ycPXypUrMWHCBIwbNw7t2rXD+vXrYWNjg40bNxptf/z4cfTs2RMjR46Ep6cngoKCMGLECIPRspCQEAwZMgStWrXCCy+8gCVLlsDOzg4nTpwAAKSnp2PDhg1YuXIl+vfvD29vb2zatAnHjx/Xt6HKOX38CbKzNKjnaolO3k5Sl0NEREREZHYspDx4QUEBEhMTMXv2bP0ymUyGgQMHIj4+3ug2/v7+2LJlCxISEuDj44Nbt25hz549CA0NNdpeo9Fg27ZtyM7Ohp+fHwAgMTERKpUKAwcO1Ldr06YNmjZtivj4eLz44osl9pOfn4/8/Hz964yMDACASqWCSqWq/MlXId3xpapDFEUc3Ft4yWbvAfWg0aih0UhSiklI3b+1HfvXtNi/psX+NS32r+mxj02L/Wta5tS/Fa1B0vCVmpoKjUaD+vXrGyyvX78+rl69anSbkSNHIjU1Fb169YIoilCr1Zg0aZLBZYcAcOHCBfj5+SEvLw92dnbYuXMn2rVrBwB49OgRLC0t4eTkVOK4jx49MnrcZcuWYeHChSWW79+/HzY25nF/U3R0tCTHfZykxMN77pBbaJGRfxp79oiS1GFqUvVvXcH+NS32r2mxf02L/Wt67GPTYv+aljn0b05OToXaSRq+nkVsbCyWLl2KtWvXwtfXFzdu3MC0adOwePFihIeH69u1bt0a586dQ3p6OrZv344xY8YgLi5OH8Aqa/bs2QgLC9O/zsjIQJMmTRAUFAQHB4fnPq/noVKpEB0djcDAQCgU1T+9+79W/QYgE359XPHysM7VfnxTk7p/azv2r2mxf02L/Wta7F/TYx+bFvvXtMypf3VXxZVH0vDl6uoKuVxeYpbBpKQkeHh4GN0mPDwcoaGhGD9+PACgY8eOyM7OxsSJEzF37lzIZIW3sVlaWsLLywsA4O3tjVOnTuHzzz/HV199BQ8PDxQUFCAtLc1g9Kus4yqVSiiVyhLLFQqF5N9sHSlqSX6Uh8vnMwEA/Qd7mE1fmII5fa9rI/avabF/TYv9a1rsX9NjH5sW+9e0zKF/K3p8SSfcsLS0hLe3N2JiYvTLtFotYmJi9PdnFZeTk6MPWDpyuRxA4b1HpdFqtfp7try9vaFQKAyOe+3aNdy9e7fU45JxcftTIIpAu84OqN/QSupyiIiIiIjMluSXHYaFhWHMmDHo3r07fHx8sGrVKmRnZ2PcuHEAgNGjR6NRo0ZYtmwZgMKZDFeuXImuXbvqLzsMDw9HSEiIPoTNnj0bgwcPRtOmTZGZmYnIyEjExsYiKioKAODo6Ih33nkHYWFhqFevHhwcHDBlyhT4+fkZnWyDjMvN0eBE3GMAQP9BfKgyEREREVFZJA9fw4cPR0pKCubPn49Hjx6hS5cu2Ldvn34Sjrt37xqMdM2bNw+CIGDevHm4f/8+3NzcEBISgiVLlujbJCcnY/To0Xj48CEcHR3RqVMnREVFITAwUN/ms88+g0wmw2uvvYb8/HwEBwdj7dq11XfitcCJw4+Rl6eFR0MrtOloL3U5RERERERmTfLwBQCTJ0/G5MmTja6LjY01eG1hYYGIiAhERESUur8NGzaUe0wrKyusWbMGa9asqVStVEirFREbVfgg7IBBbhAEQeKKiIiIiIjMm+QPWaaa6eLZdKQmF8DGVg7fXi5Sl0NEREREZPYYvuiZHNpXOOrVs78rLJV8GxERERERlYefmqnS7t/Nwa+XsyCTAX0D3aQuh4iIiIioRmD4oko7tC8FANClhxOcXSwlroaIiIiIqGZg+KJKycxQ4dTxJwCAfpxenoiIiIiowhi+qFKOxqRCrRLRrIUNmreylbocIiIiIqIag+GLKkyt1uLIgVQAQMAgd04vT0RERERUCQxfVGFnTqYhPU0FRycFuvk6SV0OEREREVGNwvBFFSKKon56+d4DXWFhwbcOEREREVFl8BM0Vcjt69m4eysHFgoBvQa4Sl0OEREREVGNw/BFFaIb9erhXw/2DgqJqyEiIiIiqnkYvqhcTx8X4NypNABAv0F8qDIRERER0bNg+KJyxUWnQKsFXmhnh0ZNbaQuh4iIiIioRmL4ojIV5Gtx7OAf08sH86HKRERERETPiuGLynTy6GPkZGvg6m6Jjt0cpS6HiIiIiKjGYviiUomiiNh9KQCAvkHukMn4UGUiIiIiomfF8EWlunohE48e5MHKSga/vi5Sl0NEREREVKMxfFGpDkUVTi//Yl8XWNvIJa6GiIiIiKhmY/gio5Ie5OHSuQwIAtA3iNPLExERERE9L4YvMip2f+G9Xh26OsLdw0riaoiIiIiIaj6GLyohJ1uNE4cfAwACgjnqRURERERUFRi+qITjsY9RkK9Fw8ZWaN3eXupyiIiIiIhqBYYvMqDRiIj745LDgEHuEAROL09EREREVBUYvsjAhTNpeJJaAFs7OXr0rCd1OUREREREtQbDFxk4uLdw1KvXAFdYWvLtQURERERUVfjpmvR+/y0HN69lQSYH+gzkRBtERERERFWJ4Yv0Du0rfKhyNx9nONWzlLgaIiIiIqLaheGLAAAZ6Sokxj8FAPQb7C5xNUREREREtQ/DFwEAjhxIhVotormXLTxb2kpdDhERERFRrcPwRVCptDhyoHCijX6DeK8XEREREZEpMHwRzpx4iswMNZycFejSw1nqcoiIiIiIaiWGrzpOFEX9RBt9g9wgt+BDlYmIiIiITIHhq467eS0bv/+WC4WlAP9+rlKXQ0RERERUazF81XG6US+fnvVgZ28hcTVERERERLUXw1cd9jglH+dPpwEAAgZxenkiIiIiIlNi+KrD4vanQBSBNh3s0bCxtdTlEBERERHVarzOrI7Ky9PgeOxjAEA/jnoRERFRFdBoNFCpVFKXYTZUKhUsLCyQl5cHjUYjdTm1TnX2r0KhgFwuf+79MHzVUScPP0FujgbuHkq06+wgdTlERERUg4miiEePHiEtLU3qUsyKKIrw8PDA77//DkHgjNJVrbr718nJCR4eHs91LIavOkirFRG7/4/p5YPdIJPxHwMiIiJ6drrg5e7uDhsbGwaNP2i1WmRlZcHOzg4yGe/2qWrV1b+iKCInJwfJyYWfnxs0aPDM+2L4qoOu/JKB5If5sLKW4cXeLlKXQ0RERDWYRqPRBy8XF36uKEqr1aKgoABWVlYMXyZQnf1rbV04P0JycjLc3d2f+RJEvgvqIN308v4BrrCyfv5rV4mIiKju0t3jZWNjI3ElRKale48/z32NDF91zMN7ubhyIROCAPQNcpO6HCIiIqoleKkh1XZV8R5n+KpjYqNSAACdvB3h6q6UuBoiIiIiorqD4asOyc5S4+RRTi9PREREZCqenp5YtWpVhdvHxsZCEATOFFlHMHzVIccOpUJVIKJxM2t4tbGTuhwiIiIiyQiCUOafBQsWPNN+T506hYkTJ1a4vb+/Px4+fAhHR8dnOt6zaNOmDZRKJR49elRtx6RCDF91hEYtIm5/4SWHAcHuvC6biIiI6rSHDx/q/6xatQoODg4Gy2bMmKFvK4oi1Gp1hfbr5uZWqclHLC0tn/vZUZVx9OhR5Obm4q9//Su+/vrrajlmWeraQ7klD19r1qyBp6cnrKys4Ovri4SEhDLbr1q1Cq1bt4a1tTWaNGmC6dOnIy8vT79+2bJl6NGjB+zt7eHu7o5hw4bh2rVrBvt49OgRQkND4eHhAVtbW3Tr1g07duwwyfmZi3On05D2RAU7Bwt093OWuhwiIiKqxURRRH6eRpI/oihWqEYPDw/9H0dHRwiCoH999epV2NvbY+/evfD29oZSqcTRo0dx8+ZNvPzyy6hfvz7s7OzQo0cPHDhwwGC/xS87dHZ2xr///W+88sorsLGxQatWrbB79279+uKXHW7evBlOTk6IiopC27ZtYWdnh0GDBuHhw4f6bdRqNaZOnQonJye4uLhg5syZGDNmDIYNG1bueW/YsAEjR45EaGgoNm7cWGL9vXv3MGLECNSrVw+2trbo3r07Tp48qV//008/oUePHrCysoKrqyteeeUV/TpBELBr1y6D/Tk5OWHz5s0AgN9++w2CIGDr1q3o27cvrKys8O233+Lx48cYMWIEGjVqBBsbG3Ts2BHfffedwX60Wi0+/fRTeHl5QalUomnTpli6dCkAYODAgZg8ebJB+5SUFFhaWiImJqbcPqlOkj7na+vWrQgLC8P69evh6+uLVatWITg4GNeuXYO7e8l7kiIjIzFr1ixs3LgR/v7++PXXXzF27FgIgoCVK1cCAOLi4vD++++jR48eUKvVmDNnDoKCgnD58mXY2toCAEaPHo20tDTs3r0brq6uiIyMxBtvvIHTp0+ja9eu1doH1SX2j+nlew9whcJS8sxNREREtVhBvhZh75yX5NgrN3SG0qpqHqUza9YsrFixAi1atICzszN+//13DBkyBEuWLIFSqcQ333yDkJAQXLt2DU2bNi11P4sXL8ann36Kv//97/jiiy8watQo3LlzB/Xq1TPaPicnBytWrMB//vMfyGQyvPXWW5gxYwa+/fZbAMD//d//4dtvv8WmTZvQtm1bfP7559i1axf69etX5vlkZmZi27ZtOHnyJNq0aYP09HQcOXIEvXv3BgBkZWWhb9++aNSoEXbv3g0PDw+cOXMGWq0WAPDzzz/jlVdewdy5c/HNN9+goKAAe/bseaZ+/cc//oGuXbvCysoKeXl58Pb2xsyZM+Hg4ICff/4ZoaGhaNmyJXx8fAAAs2fPxr/+9S989tln6NWrFx4+fIjLly8DAN5++21MnToV//jHP6BUFk4ot2XLFjRq1Aj9+/evdH2mJGn4WrlyJSZMmIBx48YBANavX4+ff/4ZGzduxKxZs0q0P378OHr27ImRI0cCKPzNwogRIwzS+L59+wy22bx5M9zd3ZGYmIg+ffro97Nu3Tr9N3PevHn47LPPkJiYWCvD152b2bh1PRtyuYDeAzm9PBEREVFFLFq0CIGBgfrX9erVQ+fOnfWvFy9ejJ07d2L37t0lRl6KGjNmDEaMGAEAWLp0KVavXo2EhAQMGjTIaHuVSoX169ejZcuWAIDJkydj0aJF+vVffPEFZs+erR91+vLLLysUgr7//nu0atUK7du3BwC8+eab2LBhgz58RUZGIiUlBadOndIHQy8vL/32S5YswZtvvomFCxfqlxXtj4r64IMP8OqrrxosK3qZ55QpUxAVFYUffvgBPj4+yMzMxOeff44vv/wSY8aMAQC0bNkS/v7+yMjIwKuvvoqpU6fiv//9L9544w0AhRlAN0hjTiQLXwUFBUhMTMTs2bP1y2QyGQYOHIj4+Hij2/j7+2PLli1ISEiAj48Pbt26hT179iA0NLTU46SnpwOAwW8W/P39sXXrVgwdOhROTk744YcfkJeXh4CAgFL3k5+fj/z8fP3rjIwMAIU/HFJfq6o7fml1xOxNAgB09XWEjW3du7b2eZXXv/R82L+mxf41LfavabF/Ta8q+lilUkEURWi1Wv0IiYUCWPGvjlVSY2VZKKCvo6J07Yv/3a1bN4N9ZWVlYeHChdizZw8ePnwItVqN3Nxc3Llzx6Cdrj90l0B27NhRv97a2hoODg549OiRQZ/pvtZqtbCxsUHz5s316+rXr4/k5GRotVqkp6cjKSkJ3bt3168XBEFfa1nnvnHjRowaNUrfZuTIkejXrx8+//xz2Nvb4+zZs+jatSucnJyM7ufcuXN45513yjyGsRqKnpuxftVoNFi2bBm2bduG+/fvo6CgAPn5+bC2toZWq8WlS5eQn5+Pfv36lehnAFAqlXjrrbewYcMG/PWvf8WZM2dw8eJF7Nq1q9LvhbLovqcqlQpyueHoakV/hiQLX6mpqdBoNKhfv77B8vr16+Pq1atGtxk5ciRSU1PRq1cv/Y2PkyZNwpw5c4y212q1+OCDD9CzZ0906NBBv/yHH37A8OHD4eLiAgsLC9jY2GDnzp0Gyb64ZcuWGaR8nf3795vNE92jo6NLLMvLleHMiYYABFjYXMWePReqv7Bawlj/UtVh/5oW+9e02L+mxf41vefpYwsLC3h4eCArKwsFBQVVWNWzyX+GEvLy8iCKov6X6zk5OQAKP0vqlgHA9OnTERsbi8WLF6N58+awtrbGmDFjkJWVpW+n1WqRl5dnsJ1arTZ4rTtGRkaG/liZmZmQyWTIy8uDhYWFQfui9emWZ2dnlzhG8XqLunr1Kk6cOIGEhASDK8w0Gg02b96MMWPGQC6XG61VR3eJYGnrBUHQn5eOSqXSb5OVlaVfXrTNZ599hi+//BJLly5Fu3btYGtri9mzZ+v3pdFoAMCgn4vKzMzE8OHD0adPH1y5cgX//Oc/0adPHzg7O5da67MoKChAbm4uDh8+XGICFt33sTySXnZYWbGxsVi6dCnWrl0LX19f3LhxA9OmTcPixYsRHh5eov3777+Pixcv4ujRowbLw8PDkZaWhgMHDsDV1RW7du3CG2+8gSNHjqBjR+O/pZk9ezbCwsL0rzMyMtCkSRMEBQXBwcGhak+0klQqFaKjoxEYGAiFQmGwbs+ORxDFFLR4wQZvvhVYyh6oLGX1Lz0/9q9psX9Ni/1rWuxf06uKPs7Ly8Pvv/8OOzs7WFlZVXGF1cPKygqCIOg/0+l+sW5vb2/wOe/06dMYN26c/haYrKws/P7777C0tNS3k8lksLKygoODg35kRjfapSMIgr5N8WMVr0W3PQA4ODjAwcEB9evXx5UrVzB48GAAhQHqwoUL6Ny5c6mfS3/44Qf06dMHX3zxhcHyzZs347vvvsOUKVPg7e2N//znP1Cr1UbvR+vcuTOOHz+O9957z+gx3NzckJ6erq/h+vXryMnJ0Z+rnV3ho45sbW0N6kxMTMTLL7+MCRMmACgMsLdv30bbtm3h4OCArl27wtraGidPnjT4rC6KIjIzM2Fvbw8/Pz90794dW7duxY4dO7B69eoq/4yel5cHa2tr9OnTp8R7vaIhT7Lw5erqCrlcjqSkJIPlSUlJ8PDwMLpNeHg4QkNDMX78eACFQ7jZ2dmYOHEi5s6dC5nsz4kkJk+ejP/97384fPgwGjdurF9+8+ZNfPnll7h48aL+etfOnTvjyJEjWLNmDdavX2/02EqlUn8DX1EKhcJs/kMoXouqQIvjsU8BAP0G1TebOmsqc/pe10bsX9Ni/5oW+9e02L+m9zx9rNFoIAgCZDKZwWexmkRXt7G/i55Tq1atsHPnTvzlL3+BIAgIDw+HVqvVn7+O7nXRS96K941u38WPVbwGY3VNmTIFy5cvR6tWrdCmTRt88cUXePr0aanfA5VKhS1btmDRokXo1KmTwboJEybgs88+w5UrVzBq1CgsX74cr776KpYtW4YGDRrg7NmzaNiwIfz8/BAREYEBAwbAy8sLb775JtRqNfbs2YOZM2cCAPr37481a9bA398fGo0GM2fOhEKhKPVcdV544QVs374dJ06cgLOzM1auXImkpCS0a9cOMpkMNjY2mDlzJmbNmgUrKyv07NkTKSkpuHDhAl5//XV9f48fPx6TJ0+Gra0tXnvttSp/P8pkMgiCYPTnpaI/P5L9hFhaWsLb29tg+ketVouYmBj4+fkZ3SYnJ6dEJ+qut9T9ZkEURUyePBk7d+7EwYMH0bx58xL7AEr+AMjl8iq9JtQcnDr+BFmZaji7WKJzdyepyyEiIiKq0VauXAlnZ2f4+/sjJCQEwcHB6NatW7XXMXPmTIwYMQKjR4+Gn58f7OzsEBwcXOrI4+7du/H48WODaeF12rZti7Zt22LDhg2wtLTE/v374e7ujiFDhqBjx45Yvny5/vN2QEAAtm3bht27d6NLly7o37+/wWOi/vGPf6BJkybo3bs3Ro4ciRkzZlTo9px58+ahW7duCA4ORkBAADw8PEpMmx8eHo4PP/wQ8+fPR9u2bTF8+HCkpKQYtBkxYgQsLCwwYsQI8x2FFSX0/fffi0qlUty8ebN4+fJlceLEiaKTk5P46NEjURRFMTQ0VJw1a5a+fUREhGhvby9+99134q1bt8T9+/eLLVu2FN944w19m/fee090dHQUY2NjxYcPH+r/5OTkiKIoigUFBaKXl5fYu3dv8eTJk+KNGzfEFStWiIIgiD///HOFa09PTxcBiOnp6VXUG8+uoKBA3LVrl1hQUKBfptVqxSWzLot/G5ko7v/pkYTV1XzG+peqDvvXtNi/psX+NS32r+lVRR/n5uaKly9fFnNzc6uwstpBo9GIT58+FTUajcmP88ILL4jz5s0z6XHMTfH+vX37tiiTycTExESTHK+s93pFs4Gk93zpEuv8+fPx6NEjdOnSBfv27dNPwnH37l2DEap58+ZBEATMmzcP9+/fh5ubG0JCQrBkyRJ9m3Xr1gFAiZkLN23ahLFjx0KhUGDPnj2YNWsWQkJCkJWVBS8vL3z99dcYMmSI6U+6mly/nIX7d3NhqZShZz8XqcshIiIioipy584d7N+/H3379kV+fj6+/PJL3L59W38vWl2jUqnw9OlTzJs3Dy+++KIko5EVJfmEG5MnTy71uQixsbEGry0sLBAREYGIiIhS9ydW4KnmrVq1wo4dOypVZ01zKKrwocq+vevBxlbybzMRERERVRGZTIbNmzdjxowZEEURHTp0wIEDB9C2bVupS5PEsWPHMGDAAP29Y+aMn8proZSkfFw4U/h8s4Bgd4mrISIiIqKq1KRJExw7dkzqMsxGQEBAhQZgzEHNnJKGyhS3PwWiCLTr5ACPhmZ6syERERERUR3D8FXL5OZoEB+bCgAIGOQmcTVERERERKTD8FXLnDj8GHl5WtRvqETbjtI+/JmIiIiIiP7E8FWLaLUiYvcXPu8gINgdMpkgcUVERERERKTD8FWLXD6fidSkfFjbyOHbq57U5RARERERUREMX7VI3P7Ce7169nOB0koucTVERERERFQUw1ctkZmmwPUr2ZDJgL5BnF6eiIiIqDoEBATggw8+0L/29PTEqlWrytxGEATs2rXruY9dVfuh6sPwVUvcvW4HAOjc3Qn1XC0lroaIiIjIvIWEhGDQoEFG1x05cgSCIOCXX36p9H5PnTqFiRMnPm95BhYsWIAuXbqUWP7w4UMMHjy4So9VmtzcXNSrVw+urq7Iz8+vlmPWRgxftUBWphoP79oAAPoN4qgXERERUXneeecdREdH4969eyXWbdq0Cd27d0enTp0qvV83NzfY2NhURYnl8vDwgFKprJZj7dixA+3bt0ebNm0kH20TRRFqtVrSGp4Vw1ctEB/7BFqNDE08rdHiBVupyyEiIqI6ThRFqESVJH9EUaxQjS+99BLc3NywefNmg+VZWVnYtm0b3nnnHTx+/BgjRoxAo0aNYGNjg44dO+K7774rc7/FLzu8efMmAgICYGVlhXbt2iE6OrrENjNnzsQLL7wAGxsbtGjRAuHh4VCpVACAzZs3Y+HChTh//jwEQYAgCPqai192eOHCBfTv3x/W1tZwcXHBxIkTkZWVpV8/duxYDBs2DCtWrECDBg3g4uKC999/X3+ssmzYsAFvvfUW3nrrLWzYsKHE+kuXLuGll16Cg4MD7O3t0bt3b9y8eVO/fuPGjWjfvj2USiUaNGiAyZMnAwB+++03CIKAc+fO6dumpaVBEATExsYCAGJjYyEIAvbu3Qtvb28olUocPXoUN2/exMiRI9GgQQPY2dmhR48eOHDggEFd+fn5mDlzJpo0aQKlUgkvLy9s2LABoijCy8sLK1asMGh/7tw5CIKAGzdulNsnz8LCJHulaqNRizga8xgA0DfIBYLA6eWJiIhIWmqosTZtrSTH/pvT36CAotx2FhYWGD16NDZv3oy5c+fqP0Nt27YNGo0GI0aMQFZWFry9vTFz5kw4ODjg559/RmhoKFq2bAkfH59yj6HVahEaGoqGDRvi5MmTSE9PN7g/TMfe3h6bN29Gw4YNceHCBUyYMAH29vb4+OOPMXz4cFy8eBH79u3TBwtHR8cS+8jOzkZwcDD8/Pxw6tQpJCcnY/z48Zg8ebJBwDx06BAaNGiAQ4cO4caNGxg+fDi6dOmCCRMmlHoeN2/eRHx8PH788UeIoojp06fjzp07aNasGQDg/v376NOnDwICAnDw4EE4ODjg2LFj+tGpdevWISwsDMuXL8fgwYORnp6OY8eOldt/xc2aNQsrVqxAixYt4OzsjDt37iAwMBDLly+HtbU1vvnmG4SEhODatWto2rQpAGD06NGIj4/H6tWr0blzZ9y+fRupqakQBAFvv/02Nm3ahBkzZuiPsWnTJvTp0wdeXl6Vrq8iGL5quPx8DTp0c8CZhCR08Sn5g0hERERExr399tv4+9//jri4OAQEBAAo/PD92muvwdHREY6OjgYfzKdMmYKoqCj88MMPFQpfBw4cwPXr17F//340btwYALB06dIS92nNmzdP/7WnpydmzJiB77//Hh9//DGsra1hZ2cHCwsLeHh4lHqsyMhI5OXl4ZtvvoGtbeGVUF9++SVCQkLwf//3f6hfvz4AwNnZGV9++SXkcjnatGmDoUOHIiYmpszwtXHjRgwePBjOzs4AgODgYGzatAkLFiwAAKxZswaOjo74/vvvoVAUBt8XXnhBv/0nn3yCDz/8ENOmTdMv69GjR7n9V9yiRYsQGBiof+3k5ITmzZvDwcEBMpkMixcvxs6dO7F7925MnjwZv/76K3744QdER0dj4MCBAIAWLVrotx87dizmz5+PhIQE+Pj4QKVSITIyssRoWFVi+KrhbGwt8ProRrCudx4WFryKlIiIiKRnAQv8zelvkh27otq0aQN/f39s3LgRAQEBuHHjBo4cOYJFixYBADQaDZYuXYoffvgB9+/fR0FBAfLz8yt8T9fVq1fRqFEjNGzYUL/Mz8+vRLutW7di9erVuHnzJrKysqBWq+Hg4FDh8wCAK1euoHPnzvrgBQA9e/aEVqvFtWvX9OGrffv2kMv/fCRRgwYNcOHChVL3q9Fo8PXXX+Pzzz/XL3vrrbcwY8YMzJ8/HzKZDOfOnUPv3r31wauo5ORkPHjwAAMGDKjU+RjTvXt3g9dZWVkIDw/HgQMH8PDhQ6jVauTm5uLu3bsACi8hlMvl6Nu3r9H9NWzYEEOHDsXGjRvh4+ODn376Cfn5+Xj99defu9bS8NN6LSHjd5KIiIjMhCAIUAgKSf5U9haMd955Bzt27EBmZiY2bdqEli1b6j+s//3vf8fnn3+OmTNn4tChQzh37hyCg4NRUFBQZX0VHx+PUaNGYciQIfjf//6Hs2fPYu7cuVV6jKKKByRBEKDVakttHxUVhfv372P48OGwsLCAhYUF3nzzTdy5cwcxMTEAAGtr61K3L2sdAMj++BBb9F690u5BKxosAeCjjz7C//73P3zyySc4cuQIzp07h44dO+r7rrxjA8D48ePx/fffIzc3F5s2bcLw4cNNOmEKP7ITERERUZ31xhtvQCaTITIyEt988w3efvttfYA7duwYXn75Zbz11lvo3LkzWrRogV9//bXC+27Tpg3u37+Phw8f6pedOHHCoM3x48fRrFkzzJ07F927d0erVq1w584dgzaWlpbQaDRlHqtt27Y4f/48srOz9cuOHTsGmUyG1q1bV7jm4jZs2IA333wT586dM/jz5ptv6ife6NSpE44cOWI0NNnb28PT01Mf1Ipzc3MDAIM+Kjr5RlmOHz+OkSNH4pVXXkHHjh3h4eGB3377Tb++Y8eO0Gq1iIuLK3UfQ4YMga2tLdatW4d9+/bh7bffrtCxnxXDFxERERHVWXZ2dhg+fDhmz56Nhw8fYuzYsfp1rVq1QnR0NI4fP44rV67g3XffRVJSUoX3PXDgQHh5eWHs2LE4f/48jhw5grlz5xq0adWqFe7evYvvv/8eN2/exOrVq7Fz506DNp6enrh9+zbOnTuH1NRUo8/ZGjVqFKysrDBmzBhcvHgRhw4dwpQpUxAaGqq/5LCyUlJS8NNPP2HMmDHo0KGDwZ/Ro0dj165dePL/7d17TJb1/8fx1y3KyUQ8Ah4yTUOl0FBTPKSGhYcsmqYWc3goU8FJ66ROU2cta5odNLKDumnLaZvOSjPUwKKTqRgaOTQyNw/YNEEUPPD5/uG8f79bAbmB67q54fnY7g2u+3PdvO9X7z98d53OnlVSUpLy8/M1btw4/fbbb8rJydHatWt1+PBhSdefU7Z06VK99957ysnJ0b59+/T+++9Lun50qk+fPlq8eLGys7OVnp7ucg1ceTp27Kgvv/xSmZmZOnDggJ5++mmXo3h33XWXEhISNGnSJG3evFm5ublKS0vThg0bnGt8fHw0YcIEzZ49W506dSr1tNDqxPAFAACAOm3y5Mk6d+6cYmNjXa7Pmjt3rqKiohQbG6tBgwYpNDRUcXFxFf7cevXqae3atSoqKtIDDzygZ555Rq+//rrLmscee0zPP/+8kpKS1L17d/3444+aN2+ey5pRo0Zp6NChGjx4sFq0aFHq7e4DAwO1fft2nT17Vr169dLo0aMVExOj5cuXuxfG/3Pj5h2lXa8VExOjgIAArVu3Ts2aNdOuXbt04cIFDRw4UD169NDHH3/sPMUxISFB77zzjj744ANFRETo0UcfVU5OjvOzVq1apatXr6pHjx5KTk7Wa6+9VqH6li5dquDgYPXv318jR45UbGysoqKiXNakpKRo9OjRmj59ujp37qxnn33W5eigdP2//+XLlzVx4kR3I3Kbw1T0YQhwkZ+fr8aNG+v8+fNuXxBZ3a5cuaKtW7dq+PDhpV7oiKohX2uRr7XI11rkay3ytV51ZFxUVKTc3Fy1b99e/v7+1VyhdyspKVF+fr7zbnyoXtWV7/fff6+YmBgdP3683KOE5fV6RWcD7nYIAAAAoM4pLi7WmTNntGDBAj355JOVPj3THYzgAAAAAOqczz//XO3atdN///2nt956y5a/yfAFAAAAoM6ZMGGCrl27pr1796p169a2/E2GLwAAAACwAcMXAAAAqox7uKG2q44eZ/gCAABApd24S+LFixc9XAlgrRs9XpW7r3K3QwAAAFSaj4+PgoODlZeXJ+n686YcDoeHq6oZSkpKdPnyZRUVFXGreQvYla8xRhcvXlReXp6Cg4Pl4+NT6c9i+AIAAECVhIaGSpJzAMN1xhhdunRJAQEBDKQWsDvf4OBgZ69XFsMXAAAAqsThcCgsLEwtW7bUlStXPF1OjXHlyhXt3r1bDz74IA8Kt4Cd+TZo0KBKR7xuYPgCAABAtfDx8amWf6DWFj4+Prp69ar8/f0Zvizgjfly8ikAAAAA2IDhCwAAAABswPAFAAAAADbgmq9KuvGQtfz8fA9Xcv1iw4sXLyo/P99rznf1JuRrLfK1Fvlai3ytRb7WI2Nrka+1alK+N2aC2z2ImeGrkgoKCiRJbdu29XAlAAAAAGqCgoICNW7cuMz3HeZ24xlKVVJSohMnTqhRo0Yef25Dfn6+2rZtq+PHjysoKMijtdRG5Gst8rUW+VqLfK1FvtYjY2uRr7VqUr7GGBUUFKhVq1blPvCZI1+VVK9ePbVp08bTZbgICgryeOPVZuRrLfK1Fvlai3ytRb7WI2Nrka+1akq+5R3xuoEbbgAAAACADRi+AAAAAMAGDF+1gJ+fn+bPny8/Pz9Pl1Irka+1yNda5Gst8rUW+VqPjK1Fvtbyxny54QYAAAAA2IAjXwAAAABgA4YvAAAAALABwxcAAAAA2IDhCwAAAABswPDlBXbv3q2RI0eqVatWcjgc2rx58233SUtLU1RUlPz8/NSxY0etWbPG8jq9lbv5pqWlyeFw3PI6deqUPQV7kTfeeEO9evVSo0aN1LJlS8XFxenw4cO33W/jxo3q3Lmz/P39dd9992nr1q02VOt9KpPvmjVrbuldf39/myr2PikpKYqMjHQ+wDM6Olrbtm0rdx/6t+LczZf+rbzFixfL4XAoOTm53HX0b+VUJF/61z0LFiy4Ja/OnTuXu4839C/DlxcoLCxUt27dtGLFigqtz83N1YgRIzR48GBlZmYqOTlZzzzzjLZv325xpd7J3XxvOHz4sE6ePOl8tWzZ0qIKvVd6eroSExP1888/KzU1VVeuXNEjjzyiwsLCMvf58ccf9dRTT2ny5Mnav3+/4uLiFBcXp4MHD9pYuXeoTL6SFBQU5NK7x44ds6li79OmTRstXrxYe/fu1W+//aaHHnpIjz/+uA4dOlTqevrXPe7mK9G/lbFnzx6tXLlSkZGR5a6jfyunovlK9K+7IiIiXPL64YcfylzrNf1r4FUkmU2bNpW75uWXXzYREREu28aOHWtiY2MtrKx2qEi+3333nZFkzp07Z0tNtUleXp6RZNLT08tcM2bMGDNixAiXbb179zbPPfec1eV5vYrku3r1atO4cWP7iqqFmjRpYj755JNS36N/q668fOlf9xUUFJhOnTqZ1NRUM3DgQDNz5swy19K/7nMnX/rXPfPnzzfdunWr8Hpv6V+OfNVCP/30k4YMGeKyLTY2Vj/99JOHKqqdunfvrrCwMD388MPKyMjwdDle4fz585Kkpk2blrmG/q28iuQrSRcuXFC7du3Utm3b2x5lwP+5du2a1q9fr8LCQkVHR5e6hv6tvIrkK9G/7kpMTNSIESNu6cvS0L/ucydfif51V05Ojlq1aqUOHTooPj5e//zzT5lrvaV/63u6AFS/U6dOKSQkxGVbSEiI8vPzdenSJQUEBHiostohLCxMH374oXr27Kni4mJ98sknGjRokH755RdFRUV5urwaq6SkRMnJyerXr5/uvffeMteV1b9cU1e+iuYbHh6uVatWKTIyUufPn9eSJUvUt29fHTp0SG3atLGxYu+RlZWl6OhoFRUV6Y477tCmTZvUtWvXUtfSv+5zJ1/61z3r16/Xvn37tGfPngqtp3/d426+9K97evfurTVr1ig8PFwnT57UwoULNWDAAB08eFCNGjW6Zb239C/DF+Cm8PBwhYeHO3/v27evjh49qmXLlmnt2rUerKxmS0xM1MGDB8s9XxuVV9F8o6OjXY4q9O3bV126dNHKlSu1aNEiq8v0SuHh4crMzNT58+f1xRdfKCEhQenp6WUOCHCPO/nSvxV3/PhxzZw5U6mpqdzUwQKVyZf+dc+wYcOcP0dGRqp3795q166dNmzYoMmTJ3uwsqph+KqFQkNDdfr0aZdtp0+fVlBQEEe9LPLAAw8wVJQjKSlJX331lXbv3n3b/7tXVv+GhoZaWaJXcyffmzVo0ED333+/jhw5YlF13s/X11cdO3aUJPXo0UN79uzRu+++q5UrV96ylv51nzv53oz+LdvevXuVl5fnckbGtWvXtHv3bi1fvlzFxcXy8fFx2Yf+rbjK5Hsz+tc9wcHBuueee8rMy1v6l2u+aqHo6Gjt3LnTZVtqamq559CjajIzMxUWFubpMmocY4ySkpK0adMm7dq1S+3bt7/tPvRvxVUm35tdu3ZNWVlZ9K8bSkpKVFxcXOp79G/VlZfvzejfssXExCgrK0uZmZnOV8+ePRUfH6/MzMxSBwP6t+Iqk+/N6F/3XLhwQUePHi0zL6/pX0/f8QO3V1BQYPbv32/2799vJJm3337b7N+/3xw7dswYY8ysWbPM+PHjnev/+usvExgYaF566SWTnZ1tVqxYYXx8fMw333zjqa9Qo7mb77Jly8zmzZtNTk6OycrKMjNnzjT16tUzO3bs8NRXqLGmTZtmGjdubNLS0szJkyedr4sXLzrXjB8/3syaNcv5e0ZGhqlfv75ZsmSJyc7ONvPnzzcNGjQwWVlZnvgKNVpl8l24cKHZvn27OXr0qNm7d68ZN26c8ff3N4cOHfLEV6jxZs2aZdLT001ubq75/fffzaxZs4zD4TDffvutMYb+rSp386V/q+bmu/HRv9XrdvnSv+554YUXTFpamsnNzTUZGRlmyJAhpnnz5iYvL88Y4739y/DlBW7c2vzmV0JCgjHGmISEBDNw4MBb9unevbvx9fU1HTp0MKtXr7a9bm/hbr5vvvmmufvuu42/v79p2rSpGTRokNm1a5dniq/hSstVkks/Dhw40Jn1DRs2bDD33HOP8fX1NREREebrr7+2t3AvUZl8k5OTzZ133ml8fX1NSEiIGT58uNm3b5/9xXuJSZMmmXbt2hlfX1/TokULExMT4xwMjKF/q8rdfOnfqrl5OKB/q9ft8qV/3TN27FgTFhZmfH19TevWrc3YsWPNkSNHnO97a/86jDHGvuNsAAAAAFA3cc0XAAAAANiA4QsAAAAAbMDwBQAAAAA2YPgCAAAAABswfAEAAACADRi+AAAAAMAGDF8AAAAAYAOGLwAAAACwAcMXAAA2cDgc2rx5s6fLAAB4EMMXAKDWmzBhghwOxy2voUOHero0AEAdUt/TBQAAYIehQ4dq9erVLtv8/Pw8VA0AoC7iyBcAoE7w8/NTaGioy6tJkyaSrp8SmJKSomHDhikgIEAdOnTQF1984bJ/VlaWHnroIQUEBKhZs2aaMmWKLly44LJm1apVioiIkJ+fn8LCwpSUlOTy/r///qsnnnhCgYGB6tSpk7Zs2eJ879y5c4qPj1eLFi0UEBCgTp063TIsAgC8G8MXAACS5s2bp1GjRunAgQOKj4/XuHHjlJ2dLUkqLCxUbGysmjRpoj179mjjxo3asWOHy3CVkpKixMRETZkyRVlZWdqyZYs6duzo8jcWLlyoMWPG6Pfff9fw4cMVHx+vs2fPOv/+H3/8oW3btik7O1spKSlq3ry5fQEAACznMMYYTxcBAICVJkyYoHXr1snf399l+5w5czRnzhw5HA5NnTpVKSkpzvf69OmjqKgoffDBB/r444/1yiuv6Pjx42rYsKEkaevWrRo5cqROnDihkJAQtW7dWhMnTtRrr71Wag0Oh0Nz587VokWLJF0f6O644w5t27ZNQ4cO1WOPPabmzZtr1apVFqUAAPA0rvkCANQJgwcPdhmuJKlp06bOn6Ojo13ei46OVmZmpiQpOztb3bp1cw5ektSvXz+VlJTo8OHDcjgcOnHihGJiYsqtITIy0vlzw4YNFRQUpLy8PEnStGnTNGrUKO3bt0+PPPKI4uLi1Ldv30p9VwBAzcTwBQCoExo2bHjLaYDVJSAgoELrGjRo4PK7w+FQSUmJJGnYsGE6duyYtm7dqtTUVMXExCgxMVFLliyp9noBAJ7BNV8AAEj6+eefb/m9S5cukqQuXbrowIEDKiwsdL6fkZGhevXqKTw8XI0aNdJdd92lnTt3VqmGFi1aKCEhQevWrdM777yjjz76qEqfBwCoWTjyBQCoE4qLi3Xq1CmXbfXr13fe1GLjxo3q2bOn+vfvr88++0y//vqrPv30U0lSfHy85s+fr4SEBC1YsEBnzpzRjBkzNH78eIWEhEiSFixYoKlTp6ply5YaNmyYCgoKlJGRoRkzZlSovldffVU9evRQRESEiouL9dVXXzmHPwBA7cDwBQCoE7755huFhYW5bAsPD9eff/4p6fqdCNevX6/p06crLCxMn3/+ubp27SpJCgwM1Pbt2zVz5kz16tVLgYGBGjVqlN5++23nZyUkJKioqEjLli3Tiy++qObNm2v06NEVrs/X11ezZ8/W33//rYCAAA0YMEDr16+vhm8OAKgpuNshAKDOczgc2rRpk+Li4jxdCgCgFuOaLwAAAACwAcMXAAAAANiAa74AAHUeZ+ADAOzAkS8AAAAAsAHDFwAAAADYgOELAAAAAGzA8AUAAAAANmD4AgAAAAAbMHwBAAAAgA0YvgAAAADABgxfAAAAAGCD/wGotRoxXCdknAAAAABJRU5ErkJggg==\n",
"text/plain": [
"<Figure size 1000x600 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Plotting the training and validation accuracy.\n",
"train_val_plot.accuracy_plot(history2a, [\"SlateBlue\", \"LightGreen\"])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<caption><span style=\"font-weight: bold;\">Figure 4</span> Training and Validation accuracy for model 2.</caption>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Figure 4 also displays underfitting. From the 2nd epoch the model's accuracy remains constant. That being said, the model reaches a high of around 84% accuracy. While this is lower than the last model, it still exceeds statistical power. I suspect that the accuracy is high due to great amount of data- the number of epochs must also bear an effect as well."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"## 5.5 The third model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Yet again, the model performed pretty well. I can no longer decrease the size of the layers nor the units so I'll try to further simplify it by decreasing the number of epochs from 5 to 3. Table 4 displays the hyperparameters / parameters that I will be using for this model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<table style=\"width: 700px\">\n",
" <caption><span style=\"font-weight: bold;\">Table 4</span> Model 3 hyperparameters / parameters.</caption>\n",
" <tr style=\"background-color: #ECE5FC;\">\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Number of Layers</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Units</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Activation</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Epochs</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Batch Size</th>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">1</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[1]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[\"sigmoid\"]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">3</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">128</td>\n",
" </tr>\n",
"</table>"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 5.5.1 Building the model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"I am using the `compile_fit_model()` function to create, compile, and fit the model."
]
},
{
"cell_type": "code",
"execution_count": 51,
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/3\n",
"11250/11250 [==============================] - 69s 6ms/step - loss: 0.3814 - accuracy: 0.8356 - val_loss: 0.3651 - val_accuracy: 0.8422\n",
"Epoch 2/3\n",
"11250/11250 [==============================] - 42s 4ms/step - loss: 0.3665 - accuracy: 0.8415 - val_loss: 0.3663 - val_accuracy: 0.8415\n",
"Epoch 3/3\n",
"11250/11250 [==============================] - 44s 4ms/step - loss: 0.3669 - accuracy: 0.8415 - val_loss: 0.3661 - val_accuracy: 0.8418\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Creating, compiling, and fitting the model\n",
"history3a = compile_fit_model(units=[1], \n",
" activation=[\"sigmoid\"], \n",
" num_of_layers=1,\n",
" epochs=3, \n",
" batch_size=128)"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 5.5.2 Plotting the training and validation loss"
]
},
{
"cell_type": "code",
"execution_count": 52,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 1000x600 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Plotting the training and validation loss.\n",
"train_val_plot.loss_plot(history3a)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<caption><span style=\"font-weight: bold;\">Figure 5</span> Training and Validation loss for model 3.</caption>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Figure 5 above shows that overfitting is occurring. This is evident due to the divergence between the training loss and validation loss. While the training loss is decreasing the validation loss is increasing. This means that this model doesn't generalise well to unseen data."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 5.5.3 Plotting the training and validation accuracy"
]
},
{
"cell_type": "code",
"execution_count": 53,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAA18AAAIjCAYAAAD80aFnAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAACfBElEQVR4nOzdd3gUVd/G8e/spjcIIfQSIHkRpAqCgBUposb6WEBpCojSBBs9VsCGKF0FG0Z5QLE8oIAoNhAQRLETQm8hlPSy2Z33D8zKkgAJZDNJuD/Xlcvs7JTfnCxm7sw5ZwzTNE1ERERERETEq2xWFyAiIiIiInI+UPgSEREREREpBQpfIiIiIiIipUDhS0REREREpBQofImIiIiIiJQChS8REREREZFSoPAlIiIiIiJSChS+RERERERESoHCl4iIiIiISClQ+BIRKUP69etHVFTUWW37+OOPYxhGyRZUxuzYsQPDMHjzzTdL/diGYfD444+7X7/55psYhsGOHTvOuG1UVBT9+vUr0XrO5bMiIiLWUPgSESkCwzCK9LV69WqrSz3vDR8+HMMwSEhIOOU648aNwzAMfvnll1KsrPj27dvH448/zubNm60upVB//PEHhmEQEBDAsWPHrC5HRKTMU/gSESmCd955x+Ora9euhS5v0qTJOR3ntdde46+//jqrbcePH09WVtY5Hb8iuOuuuwCIj48/5TrvvfcezZs3p0WLFmd9nN69e5OVlUX9+vXPeh9nsm/fPp544olCw9e5fFZKyoIFC6hRowYAixcvtrQWEZHywMfqAkREyoO7777b4/UPP/zAypUrCyw/WWZmJkFBQUU+jq+v71nVB+Dj44OPj/633r59e6Kjo3nvvfeYOHFigffXrl3L9u3bmTJlyjkdx263Y7fbz2kf5+JcPislwTRN4uPj6dWrF9u3b+fdd99lwIABltZ0KhkZGQQHB1tdhoiI7nyJiJSUK6+8kmbNmrFx40Yuv/xygoKCGDt2LAAff/wx1113HbVq1cLf359GjRrx1FNP4XQ6PfZx8jie/DFOL7zwAq+++iqNGjXC39+fiy++mA0bNnhsW9iYL8MwGDp0KB999BHNmjXD39+fCy+8kM8//7xA/atXr6Zt27YEBATQqFEj5s6dW+RxZN9++y233XYb9erVw9/fn7p16zJy5MgCd+L69etHSEgIe/fu5aabbiIkJITIyEgefvjhAm1x7Ngx+vXrR6VKlahcuTJ9+/Ytcte2u+66iz///JNNmzYVeC8+Ph7DMOjZsye5ublMnDiRNm3aUKlSJYKDg7nsssv46quvzniMwsZ8mabJ008/TZ06dQgKCuKqq67it99+K7DtkSNHePjhh2nevDkhISGEhYXRo0cPfv75Z/c6q1ev5uKLLwagf//+7q6t+ePdChvzlZGRwUMPPUTdunXx9/encePGvPDCC5im6bFecT4Xp/L999+zY8cO7rzzTu68806++eYb9uzZU2A9l8vFyy+/TPPmzQkICCAyMpJrrrmGH3/80WO9BQsW0K5dO4KCgggPD+fyyy9nxYoVHjWfOOYu38nj6fJ/Ll9//TUPPPAA1apVo06dOgDs3LmTBx54gMaNGxMYGEhERAS33XZboeP2jh07xsiRI4mKisLf3586derQp08fkpOTSU9PJzg4mBEjRhTYbs+ePdjtdiZPnlzElhSR84n+RCoiUoIOHz5Mjx49uPPOO7n77rupXr06cPyCMCQkhFGjRhESEsKXX37JxIkTSU1N5fnnnz/jfuPj40lLS+O+++7DMAyee+45brnlFhITE894B+S7777jww8/5IEHHiA0NJRXXnmFW2+9lV27dhEREQHATz/9xDXXXEPNmjV54okncDqdPPnkk0RGRhbpvBctWkRmZib3338/ERERrF+/nunTp7Nnzx4WLVrksa7T6aR79+60b9+eF154gS+++IIXX3yRRo0acf/99wPHQ8yNN97Id999x+DBg2nSpAlLliyhb9++Rarnrrvu4oknniA+Pp6LLrrI49j//e9/ueyyy6hXrx7Jycm8/vrr9OzZk4EDB5KWlsa8efPo3r0769evp1WrVkU6Xr6JEyfy9NNPc+2113LttdeyadMmunXrRm5ursd6iYmJfPTRR9x22200aNCAgwcPMnfuXK644gp+//13atWqRZMmTXjyySeZOHEigwYN4rLLLgOgY8eOhR7bNE1uuOEGvvrqK+69915atWrF8uXLeeSRR9i7dy8vvfSSx/pF+VyczrvvvkujRo24+OKLadasGUFBQbz33ns88sgjHuvde++9vPnmm/To0YMBAwaQl5fHt99+yw8//EDbtm0BeOKJJ3j88cfp2LEjTz75JH5+fqxbt44vv/ySbt26Fbn9T/TAAw8QGRnJxIkTycjIAGDDhg2sWbOGO++8kzp16rBjxw5mz57NlVdeye+//+6+S52ens5ll13GH3/8wT333MNFF11EcnIyn3zyCXv27KFVq1bcfPPNLFy4kKlTp3rcAX3vvfcwTdPd/VVExIMpIiLFNmTIEPPk/4VeccUVJmDOmTOnwPqZmZkFlt13331mUFCQmZ2d7V7Wt29fs379+u7X27dvNwEzIiLCPHLkiHv5xx9/bALmp59+6l4WFxdXoCbA9PPzMxMSEtzLfv75ZxMwp0+f7l4WGxtrBgUFmXv37nUv27p1q+nj41Ngn4Up7PwmT55sGoZh7ty50+P8APPJJ5/0WLd169ZmmzZt3K8/+ugjEzCfe+4597K8vDzzsssuMwHzjTfeOGNNF198sVmnTh3T6XS6l33++ecmYM6dO9e9z5ycHI/tjh49alavXt285557PJYDZlxcnPv1G2+8YQLm9u3bTdM0zaSkJNPPz8+87rrrTJfL5V5v7NixJmD27dvXvSw7O9ujLtM8/rP29/f3aJsNGzac8nxP/qzkt9nTTz/tsd5//vMf0zAMj89AUT8Xp5Kbm2tGRESY48aNcy/r1auX2bJlS4/1vvzySxMwhw8fXmAf+W20detW02azmTfffHOBNjmxHU9u/3z169f3aNv8n8ull15q5uXleaxb2Od07dq1JmC+/fbb7mUTJ040AfPDDz88Zd3Lly83AfOzzz7zeL9FixbmFVdcUWA7ERHTNE11OxQRKUH+/v7079+/wPLAwED392lpaSQnJ3PZZZeRmZnJn3/+ecb93nHHHYSHh7tf598FSUxMPOO2Xbp0oVGjRu7XLVq0ICwszL2t0+nkiy++4KabbqJWrVru9aKjo+nRo8cZ9w+e55eRkUFycjIdO3bENE1++umnAusPHjzY4/Vll13mcS7Lli3Dx8fHfScMjo+xGjZsWJHqgePj9Pbs2cM333zjXhYfH4+fnx+33Xabe59+fn7A8e5xR44cIS8vj7Zt2xbaZfF0vvjiC3Jzcxk2bJhHV80HH3ywwLr+/v7YbMd/BTudTg4fPkxISAiNGzcu9nHzLVu2DLvdzvDhwz2WP/TQQ5imyWeffeax/Eyfi9P57LPPOHz4MD179nQv69mzJz///LNHN8sPPvgAwzCIi4srsI/8Nvroo49wuVxMnDjR3SYnr3M2Bg4cWGBM3omfU4fDweHDh4mOjqZy5coe7f7BBx/QsmVLbr755lPW3aVLF2rVqsW7777rfu/XX3/ll19+OeNYUBE5fyl8iYiUoNq1a7sv5k/022+/cfPNN1OpUiXCwsKIjIx0X6ClpKSccb/16tXzeJ0fxI4ePVrsbfO3z982KSmJrKwsoqOjC6xX2LLC7Nq1i379+lGlShX3OK4rrrgCKHh++eN+TlUPHB+bU7NmTUJCQjzWa9y4cZHqAbjzzjux2+3uWQ+zs7NZsmQJPXr08Aiyb731Fi1atCAgIICIiAgiIyNZunRpkX4uJ9q5cycAMTExHssjIyM9jgfHg95LL71ETEwM/v7+VK1alcjISH755ZdiH/fE49eqVYvQ0FCP5fkzcObXl+9Mn4vTWbBgAQ0aNMDf35+EhAQSEhJo1KgRQUFBHmFk27Zt1KpViypVqpxyX9u2bcNms9G0adMzHrc4GjRoUGBZVlYWEydOdI+Jy2/3Y8eOebT7tm3baNas2Wn3b7PZuOuuu/joo4/IzMwEjnfFDAgIcId7EZGTKXyJiJSgE/+ynu/YsWNcccUV/Pzzzzz55JN8+umnrFy5kmeffRY4fiF+JqeaVc88aSKFkt62KJxOJ127dmXp0qU89thjfPTRR6xcudI9McTJ51daMwRWq1aNrl278sEHH+BwOPj0009JS0vzGIuzYMEC+vXrR6NGjZg3bx6ff/45K1eupHPnzkX6uZytSZMmMWrUKC6//HIWLFjA8uXLWblyJRdeeKFXj3uis/1cpKam8umnn7J9+3ZiYmLcX02bNiUzM5P4+PgS+2wVxckTteQr7N/isGHDeOaZZ7j99tv573//y4oVK1i5ciURERFn1e59+vQhPT2djz76yD374/XXX0+lSpWKvS8ROT9owg0RES9bvXo1hw8f5sMPP+Tyyy93L9++fbuFVf2rWrVqBAQEFPpQ4tM9qDjfli1b+Pvvv3nrrbfo06ePe/nKlSvPuqb69euzatUq0tPTPe5+Ffe5VnfddReff/45n332GfHx8YSFhREbG+t+f/HixTRs2JAPP/zQo4tbYd3kilIzwNatW2nYsKF7+aFDhwrcTVq8eDFXXXUV8+bN81h+7Ngxqlat6n5dnG539evX54svviAtLc3j7ld+t9aSeh7Zhx9+SHZ2NrNnz/aoFY7/fMaPH8/333/PpZdeSqNGjVi+fDlHjhw55d2vRo0a4XK5+P333087wUl4eHiB2S5zc3PZv39/kWtfvHgxffv25cUXX3Qvy87OLrDfRo0a8euvv55xf82aNaN169a8++671KlTh127djF9+vQi1yMi5x/d+RIR8bL8Owwn3g3Izc1l1qxZVpXkwW6306VLFz766CP27dvnXp6QkFBgnNCptgfP8zNNk5dffvmsa7r22mvJy8tj9uzZ7mVOp7PYF7Y33XQTQUFBzJo1i88++4xbbrmFgICA09a+bt061q5dW+yau3Tpgq+vL9OnT/fY37Rp0wqsa7fbC9wdWrRoEXv37vVYlv9sqqJMsX/ttdfidDqZMWOGx/KXXnoJwzCKPH7vTBYsWEDDhg0ZPHgw//nPfzy+Hn74YUJCQtxdD2+99VZM0+SJJ54osJ/887/pppuw2Ww8+eSTBe4+ndhGjRo18hi/B/Dqq6+e8s5XYQpr9+nTpxfYx6233srPP//MkiVLTll3vt69e7NixQqmTZtGREREibWziFRMuvMlIuJlHTt2JDw8nL59+zJ8+HAMw+Cdd94p1a5ZZ/L444+zYsUKOnXqxP333+++iG/WrBmbN28+7bYXXHABjRo14uGHH2bv3r2EhYXxwQcfFGns0KnExsbSqVMnRo8ezY4dO2jatCkffvhhscdDhYSEcNNNN7nHfZ08/ff111/Phx9+yM0338x1113H9u3bmTNnDk2bNiU9Pb1Yx8p/XtnkyZO5/vrrufbaa/npp5/47LPPCtwhuv7663nyySfp378/HTt2ZMuWLbz77rsed8zgeOCoXLkyc+bMITQ0lODgYNq3b1/oeKbY2Fiuuuoqxo0bx44dO2jZsiUrVqzg448/5sEHH/SYXONs7du3j6+++qrApB75/P396d69O4sWLeKVV17hqquuonfv3rzyyits3bqVa665BpfLxbfffstVV13F0KFDiY6OZty4cTz11FNcdtll3HLLLfj7+7NhwwZq1arlfl7WgAEDGDx4MLfeeitdu3bl559/Zvny5QXa9nSuv/563nnnHSpVqkTTpk1Zu3YtX3zxRYGp9R955BEWL17Mbbfdxj333EObNm04cuQIn3zyCXPmzKFly5budXv16sWjjz7KkiVLuP/++y1/+LWIlG268yUi4mURERH873//o2bNmowfP54XXniBrl278txzz1ldmlubNm347LPPCA8PZ8KECcybN48nn3ySq6++2uNOUWF8fX359NNPadWqFZMnT+aJJ54gJiaGt99++6zrsdlsfPLJJ9x1110sWLCAcePGUbt2bd56661i7ys/cNWsWZPOnTt7vNevXz8mTZrEzz//zPDhw1m+fDkLFixwP3+quJ5++mmeeOIJfvrpJx555BG2bdvGihUr3Hew8o0dO5aHHnqI5cuXM2LECDZt2sTSpUupW7eux3q+vr689dZb2O12Bg8eTM+ePfn6668LPXZ+mz344IP873//48EHH+T333/n+eefZ+rUqWd1Pid7//33cblcHl03TxYbG8vhw4fdd03feOMNnn/+ebZv384jjzzCpEmTyMrK8nhe2ZNPPsn8+fPJyspi3LhxTJw4kZ07d3L11Ve71xk4cCCPPfYY33zzDQ899BDbt29n5cqVBdr2dF5++WX69OnDu+++y0MPPcT+/fv54osvCkzsEhISwrfffsv999/PsmXLGD58OLNmzaJx48buBzbnq169uvtZZL179y5yLSJyfjLMsvSnVxERKVNuuukmfvvtN7Zu3Wp1KSJl1s0338yWLVuKNEZSRM5vuvMlIiLA8Wm4T7R161aWLVvGlVdeaU1BIuXA/v37Wbp0qe56iUiR6M6XiIgAx7vl9evXj4YNG7Jz505mz55NTk4OP/30U4FnV4mc77Zv387333/P66+/zoYNG9i2bRs1atSwuiwRKeM04YaIiABwzTXX8N5773HgwAH8/f3p0KEDkyZNUvASKcTXX39N//79qVevHm+99ZaCl4gUie58iYiIiIiIlAKN+RIRERERESkFCl8iIiIiIiKlQGO+zpLL5WLfvn2EhoZiGIbV5YiIiIiIiEVM0yQtLY1atWphs536/pbC11nat29fgYdhioiIiIjI+Wv37t0FHsZ+IoWvsxQaGgocb+CwsDBLa3E4HKxYsYJu3brh6+traS0VkdrXu9S+3qX29S61r3epfb1Pbexdal/vKkvtm5qaSt26dd0Z4VQUvs5SflfDsLCwMhG+goKCCAsLs/yDVxGpfb1L7etdal/vUvt6l9rX+9TG3qX29a6y2L5nGo6kCTdERERERERKgcKXiIiIiIhIKVD4EhERERERKQUKXyIiIiIiIqVA4UtERERERKQUKHyJiIiIiIiUAoUvERERERGRUqDwJSIiIiIiUgoUvkREREREREqBwpeIiIiIiEgpUPgSEREREREpBQpfIiIiIiIipUDhS0REREREpBQofImIiIiIiJQChS8REREREZFSoPAlIiIiIiJSChS+KoBtedvIDM/kqOsoeWae1eWIiIiIiEghfKwuQM7ddznfkdE8g/cy34NMCDaCCbOFEWYPo5Kt0vHvbce/D7GFYDOUuUVERERESpvCVzlnmibVbNXYk7IHQsCBgwwzgwxnBvud+wusb2AQagt1h7GTQ1qQEYRhGBaciYiIiIhIxabwVc4ZhkGPwB4s+2oZPXr0wOnjJMWVQqorlVRXKinOf79Pc6XhxOl+vYc9Bfbng4/7Tllhd878bf4WnKWIiIiISPmn8FWBGIZBoC2QQFsgNahR4H3TNEk300l1/hPMTghpqa5U0l3p5JHHEdcRjriOQCHDx/wN/3/DWf7dM/u/r30N31I4UxERERGR8kfh6zxiGAahRiihtlBqU7vA+07TSZor7d+7Zq4Uj6CWZWaRY+ZwyHmIQ85DhR4jyAj6N5jZC443sxt2b5+miIiIiEiZpPAlbnbDTmV7ZSrbKxf6vsN0FOjKeGJIyyWXTDOTTGcmB5wHwOG5vYFBiC2kQFfG/DtnwUawxpuJiIiISIWl8CVF5mv4EmGPIMIeUeA90zTJMXM8x5udcOcs1ZWKk+N31tJcaYXu34799OPNDH+FMxEREREptxS+pEQYhkGAEUCALYDqVC/wvmmaZJgZx4OYM7VASEt3pePEyVHXUY66jhY63swPvwKh7MTujRpvJiIiIiJlmcKXlArDMAgxQgixhVDLp1aB952mk3RX+ilnasw0M8kll2RnMsnO5EKPEWgEFjp9fpgtjFBbqMabiYiIiIilFL6kTLAbdirZK1HJXqnQ9/PHm53qzlmumUuWmUWWM+u04808xpqdENI03kxEREREvE3hS8oFj/FmhfQuzHZlF5g+/8Q7ZyeON9vL3gLb27ETagv1CGX5QS3IDMLELIWzFBEREZGKrEyEr5kzZ/L8889z4MABWrZsyfTp02nXrt0p1582bRqzZ89m165dVK1alf/85z9MnjyZgICAAutOmTKFMWPGMGLECKZNmwbAkSNHiIuLY8WKFezatYvIyEhuuukmnnrqKSpVKvzOi5RtAbbj482qUa3Ae6ZpkmlmFjp9/okPnz7mOsYx17FC9290NHg/830q2ysXOlOjn+Hn5TMUERERkfLO8vC1cOFCRo0axZw5c2jfvj3Tpk2je/fu/PXXX1SrVvBCOj4+ntGjRzN//nw6duzI33//Tb9+/TAMg6lTp3qsu2HDBubOnUuLFi08lu/bt499+/bxwgsv0LRpU3bu3MngwYPZt28fixcv9ur5SukzDINgI5hgWzA1qVngfZfpKjje7ISQlmFmYPqY/z58uhCBRmChk4Dkf2m8mYiIiIhYHr6mTp3KwIED6d+/PwBz5sxh6dKlzJ8/n9GjRxdYf82aNXTq1IlevXoBEBUVRc+ePVm3bp3Heunp6dx111289tprPP300x7vNWvWjA8++MD9ulGjRjzzzDPcfffd5OXl4eNjebNIKbIZtuN3sOxhhb6flZvF/1b/j9aXtSbDyCjwjLMcM8c93uyg82Ch+wgxQgqdPj/Mfvz5ZjbD5s1TFBEREZEywNKUkZuby8aNGxkzZox7mc1mo0uXLqxdu7bQbTp27MiCBQtYv3497dq1IzExkWXLltG7d2+P9YYMGcJ1111Hly5dCoSvwqSkpBAWFnbK4JWTk0NOTo77dWpqKgAOhwOHw1HoNqUl//hW11FRmXkmfll+1DJr4evjCyfdxMoxc0hzpZFqHg9kaWba8XBmHu/SmEce6WY66Xnp7GNfgf3bsBFqhBJqCyXM+Gd2RiPUHdICCKjQk4Ho8+tdal/vUvt6l9rX+9TG3qX29a6y1L5FrcHS8JWcnIzT6aR6dc/nQlWvXp0///yz0G169epFcnIyl156KaZpkpeXx+DBgxk7dqx7nffff59NmzaxYcOGItfx1FNPMWjQoFOuM3nyZJ544okCy1esWEFQUFCRjuNtK1eutLqECq047WtgUIlKhBGGy9eFI8BBXkBewS//PFw2FylmCinOlML35TTwyfbx+PLN9nV/b3NWjLtm+vx6l9rXu9S+3qX29T61sXepfb2rLLRvZmZmkdYrd/3rVq9ezaRJk5g1axbt27cnISGBESNG8NRTTzFhwgR2797NiBEjWLlyZaETcJwsNTWV6667jqZNm/L444+fcr0xY8YwatQoj+3q1q1Lt27dCAsrvLtaaXE4HKxcuZKuXbvi66sHDZc0b7avy3T9+/Dpf+6UnXgHLcPMwLSbOIIdOIIL/4tKAAHH75rZwggzwgrcQSvr4830+fUuta93qX29S+3rfWpj71L7eldZat/8XnFnYmn4qlq1Kna7nYMHPcfJHDx4kBo1ahS6zYQJE+jduzcDBgwAoHnz5mRkZDBo0CDGjRvHxo0bSUpK4qKLLnJv43Q6+eabb5gxYwY5OTnY7ccvRtPS0rjmmmsIDQ1lyZIlp/2h+fv74+/vX2C5r6+v5T/sfGWplorIW+3rjz9VqFLoe3lmHmmutH8nAznpGWfZZjbZZJPtyuaQ61Ch+8gfb1bYM85CjJAyM95Mn1/vUvt6l9rXu9S+3qc29i61r3eVhfYt6vEtDV9+fn60adOGVatWcdNNNwHgcrlYtWoVQ4cOLXSbzMxMbDbPi8X8MGWaJldffTVbtmzxeL9///5ccMEFPPbYY+51U1NT6d69O/7+/nzyySdFuksmUtp8DB/C7eGE28MLfT/HzHHPyljYTI0OHGceb2YLPeVMjUFGUIUebyYiIiJSmizvdjhq1Cj69u1L27ZtadeuHdOmTSMjI8M9+2GfPn2oXbs2kydPBiA2NpapU6fSunVrd7fDCRMmEBsbi91uJzQ0lGbNmnkcIzg4mIiICPfy1NRUunXrRmZmJgsWLCA1NdV9qzAyMtId0ETKOn/Dn0ifSCKJLPCeaZpkmVmFhrIUVwpprjRcuEhxpZDiKny8mQ8+7rtkhc3U6G8UvBssIiIiIoWzPHzdcccdHDp0iIkTJ3LgwAFatWrF559/7p6EY9euXR53usaPH49hGIwfP569e/cSGRlJbGwszzzzTJGPuWnTJvfU9NHR0R7vbd++naioqHM/MRGLGYZBkBFEkC2IGhTsxuseb3ZSV8b8kJZuppNH3mmfb+Zv+HuEspPvnPkYlv8vRkRERKTMKBNXRkOHDj1lN8PVq1d7vPbx8SEuLo64uLgi7//kfVx55ZWYplncMkUqFJvx7xT3tald4P388WbuUOZM8ejemGVmkWPmkORMIsmZVOgxgo3gU945C7GVnfFmIiIiIqWhTIQvESl7zjTeLNfMLXQSkPw7Zw4cZJgZZDgz2O/cX2B7GzZCbCGEGqGkxKTwY+6PhJvh7pCm8WYiIiJS0Sh8ichZ8TP8qGqvSlV71QLvmaZJtpntGcpOuHOW5krDifP4a1KhJqzPXQ+5/+7DPd7sFHfO/G0abyYiIiLli8KXiJQ4wzAINAIJtAUWOt7MNE3SzXRSnakccRxh01+bqB5dnXTSSXGmFBxvllfwGP6Gf6HT5+d/r/FmIiIiUtbo6kRESp1hGO7xZtWoRuKuRK5udrX7GRlO0+n5fLN/ujfmd2vMH292yHmIQ87Cn28WZAS5Z2U8OaSF2kI13kxERERKncKXiJQ5dsNOZXtlKtsrF/q+w3QUOglI/nizXHLJNDPJdGYWOt7MwCj4fLMTglqwEazxZiIiIlLiFL5EpNzxNXyJsEcQYY8o8J5pmuSYOYVOApL/2j3ezJVa6P7t2E8/3szwVzgTERE5R07TSa6Zi8N0kEvuv9+bhXyPo8CyXFcujhiH1adRLApfIlKhGIZBgBFAgC2A6lQv8L5pmsefb3aKmRrTXek4cXLUdZSjrqOFjjfzw69AKDvxGWe+hm8pnKmIiEjpMk2TXE4KQEUISe7vT9jWYTpw4jznmgICAkrgzEqPwpeInFcMwyDECCHEFkItn1oF3neaTtJd6aecqTHTzCSXXJKdySQ7kws9RqARWOgkIPnjzeyG3dunKSIigmma5JHnEYDO9i6Tw3TgwDt3mezY8TP88DP88DV8j/8X3wLL3N9z/Hub08aGhA1Q1ytleYXCl4jICeyGnUr2SlSyVyr0/fzxZqe6c5Zr5pJlZpHlzOKA8wAn/54yMAixhZxypkaNNxMROb+da1e8k+8ymZglXqMN27+BiBMC0ylC0pneP9tJsBwOBz9n/VzCZ+ddCl8iIsXgMd6skN6F2a7sf8eZFXLnzMnxmRzTXGnsZW+B7e3YCbWFnvLOWYARoHAmIlKGuEzXKUNQVl4WqTVT2ZS7CWees9S64hWmJEJS/nI7dv0uOksKXyIiJSjAdny8WTWqFXjPNE0yzUyPCUBODGn5D58+5jrGMdexQvefP96ssDtnYbYw/Aw/L5+hiEj5dmJXvLMdr3Ti8rzCBgefKAZ+yP2h2HX64HPOISn/e198FZbKCIUvEZFSYhgGwUYwwbbgQsebuUwXaa60AqEsP6hlmBlFGm924tT5wWYwWeFZHHMdo4pZRePNRKRcOudZ8U4c84T3uuKdHIJ88OHwgcPUq1kPf7v/GUPSid349DzKiknhS0SkjLAZNvd4s7qFjB7OM/MK7cqYH9ZyzBz3eLODzoP/jjdrDvGZ8ZAJIUZIodPnh9mPP99Mv+xFpCScriueVbPiFcbbXfEcDgfL/lhG5wad8fXVTLii8CUiUm74GD5UsVehir1KoePNcsycApOAHMs7xr7UfZjBx7vZpJvppOels499Bba3YTvteLNAI1DdVkQqKNM0zzkkFasr3llSVzwp7xS+REQqCH/Dn0ifSCKJdC9zOBws+2oZPXr0wOHjOOWdszRXGi5cpLhSSHGlFLp/X3wLDWX5zzjTeDOR0pVn5p1zSMo1c8nsmMmcjDml1hWvuCFJXfGkIlH4EhE5DxiGQbDt+HizmtQs8L7LdJHuSj/lTI0ZZgYOHBx2Heaw63ChxwgwAgqdPj//+WY+hn7lyPmtKF3xThWSCgtRLlwlU9hJ/zTPJSSd/Gwm/bsX8aR/ESIigs2wHZ8x0R5GHeoUeD/PzCPNlfZvMDupe2O2mX38y5lNkjOp0GPkjzcrbKbGECNEf9GWMqfUZ8U7S8XtindiSLI5baz5eg1dr+xKkF8QPvioK56IFyl8iYjIGfkYPoTbwwm3hxf6fv54s8JmakxxpRR5vNmp7pxpvJkUVUl0xbNiVjyruuI5cOCb7UuQLQhfQxNCiHibwpeIiJyzwsab5TNNkywz699gdlJIO3m82W52F9iHDz7uWRkLu3Pmb/iXxmmKF7hM1ymnDz/5AbWl2hXvJOqKJyIlQf/6RUTEqwzDIMgIIsgWRA1qFHjfZbrIMDMKdGXMD2rpZjp55BVpvNmJU+efOCGILnZLjmmaOHCc1Z2kwt735gNqzzUk5X+vrngiUlL020hERCxlM2yEGqGE2kKpTe0C7+ePNzvVTI1ZZtYZx5sFG8GFztRYyVaJEFvFH29Wkl3xcsn1So127J53lfB8QG2APeCMIenE9yv6z1TKBtM0MU3++TIxXWACpuv4cofDiSPXICM9Dx8f49/1XWDyz/pmIfvJ/95l/rO/06xTyHFPuc8z7AcTXGbh+zn+XuH1g4kr//t/1snfD/DPe8e/N13/7se9T9e/7XniflxmwX26v3eBy+kkNTMUri3VH/s5UfgSEZEy7UzjzXLN3EInAcm/c+bAQYaZQYYzg/3O/QW2t2EjxBZyyvFmQUZQqd/1OF1XvLO5y+SNrngGhjsEeasrXkk+oPacLlYr8EVyXp6TxD9DWZGXhM1m89pFcn7N5j/74YSaXfn7dJn/bFf4uR9f7yx+Fqf6ueT/LMwi7PMUP5eiqcNXH/1xTp9fObWIGuWr27nCl4iIlGt+hh9V7VWpaq9a4D3TNMk2swudPj9/vJkTp/v1HvYU2IcPPv92YTzpzpm/I5DDB/3Z8lMKpt0kz3B4fDltuTiNvBNeO8izHf+v05aHK/97ex5OmwOX/fhy0+70SlsZeXaMPJ/jXw4f+Oe/Rp4POPK/7Bi5J7zOPb6MXB9cuXYMhw9mjh2Xw47LNMgyTTKLfJHsxDTzMM2sIl0ku1wmWVk1WbfyD8AohYvk81VlEn49aHUR5xXDyP8yMGxgAIbN8Fx+8ve243/0cH9f2Dr/fI8BNqPwfR5/79/9gIHNxj/vGce3++d744T9AP+sd/x7w3Z8P5ywf9s/N5wN2/H9cMJ52U7c5z/H/fe8TlzPc5/u+m3GP+v9u9zlcrJj16+l9nMrCQpfIiJSYRmGQaARSKAtsNDxZqZpkm6mFzpTY4ozxT3e7IjrCEdcRyhseJLrRhvf+PyFN3q5uRzG8aCTbceVY8eV7fPPf+1n/K+Z41NgOea53sFz/vNVmnzIzvTOFO1FkX9xeMoLwpMvDs+0Tv7FtvHPhbTtNBeZp7jgPOU+KdoF/IkX26ZpsmfPburVq4vdbve4aId/L7ZPvmh3f59/sW2c4QL+pIvtE9vzjBfwpzgX9z4La8MzXLSfaxue8md30nHzHA6WL/+cHtf2wM/P999wJCXC4XCwbFmW1WUUi8KXiIictwzDOO14M6fp9Hy+2T/dG485U0jKOAqBudj8/u3SZ7rwuFtkOP69g5T//fG7TnbI88Vw+GDLs2Pk+brvSNmcvv/81webaT/DBfw/y+xgBBsYIRR+oXiWF5yefyH3vIAFCr3YLuwv5O7vT7jYzr9o97iAt+Xv5/i6Lmcea9Z8T6dLL8XX18fjYvvE9Y5fiBv//ExP+P6f4554AW87ISycPgScHxfJxy9ef+Haazucc9dOKcgwbNjsYLcb7s+lnN8UvkRERE7BbtipbK9MZXtl9zKXy2T+jO38ve4Y/iEuml25l1tvuVIPqPUCh8PBr386qBsVqGAgIhWCpgISEREpItM0iX99Fz+tO4aPj8E9gxsSEYT7AbUKXiIicjoKXyIiIkVgmiYfvruXtV8fxjCg/9AoGjcLtbosEREpRxS+REREiuCzJQf48rPjzxG7e1B9Wl0cbnFFIiJS3ih8iYiInMFXnyex9IPjzwi7rU8dLrk8wuKKRESkPFL4EhEROY21Xx9m8TvHn/91/X9qcmX3ahZXJCIi5ZXCl4iIyClsWneUd1/bCcDV11bjmpsKPitMRESkqBS+RERECvH7L6m8OXMHpgkdr4zg5l61NZuhiIicE4UvERGRkyT8lc6rL23D6TRpc0k4Pe+tp+AlIiLnTOFLRETkBLu2ZzL7+QQcuSYXtgqjz/31sdkUvERE5NwpfImIiPzjwL5sZj6bQHaWi+gLQhgwvCE+PvpVKSIiJUO/UURERIDDh3KYPmkr6Wl51GsQxOCHGuHnr1+TIiJScvRbRUREznspxxxMn5zAsaMOatQOYMhj0QQG2a0uS0REKhiFLxEROa9lpOcxY/JWDh3MISLSj2GjowkJ9bG6LBERqYAUvkRE5LyVneVk1nMJ7NuTTaXKvgwfG0PlKn5WlyUiIhWUwpeIiJyXHLku5r6UyI5tmQSH2Bk6Jpqq1fytLktERCowhS8RETnvOPNM5k3fzt+/pREQYGPIo9HUqhNodVkiIlLBKXyJiMh5xeUyeefVHWzZlIKvr8HghxtRv1Gw1WWJiMh5QOFLRETOG6Zp8t+3drPh+6PY7DBgRENimoRaXZaIiJwnFL5EROS88cl/9/HtF8kYBvS9P4pmrStZXZKIiJxHFL5EROS8sOLTA6z45CAAd95Tj7YdqlhckYiInG8UvkREpML79otDfPz+PgBu7lWbSztXtbgiERE5Hyl8iYhIhbbh+yMsfHM3ANfcWIMu11W3uCIRETlfKXyJiEiFtWXTMd6eswPThCu6RXL9bTWtLklERM5jCl8iIlIh/fVbGq+/sh2XC9pfVoX/9K6DYRhWlyUiIucxhS8REalwtidkMOfFbeQ5TFq2rcRdA+tjsyl4iYiItRS+RESkQtm7K4tZzyWQm+Pigmah9B/aALtdwUtERKxnefiaOXMmUVFRBAQE0L59e9avX3/a9adNm0bjxo0JDAykbt26jBw5kuzs7ELXnTJlCoZh8OCDD3osf/XVV7nyyisJCwvDMAyOHTtWQmcjIiJWSjqQzYxnt5KZ4aRBTDCDRjbE19fyX3UiIiKAxeFr4cKFjBo1iri4ODZt2kTLli3p3r07SUlJha4fHx/P6NGjiYuL448//mDevHksXLiQsWPHFlh3w4YNzJ07lxYtWhR4LzMzk2uuuabQ7UREpHw6ejiX6ZMTSD2WR+16gTzwSCP8A+xWlyUiIuJmafiaOnUqAwcOpH///jRt2pQ5c+YQFBTE/PnzC11/zZo1dOrUiV69ehEVFUW3bt3o2bNngbtl6enp3HXXXbz22muEh4cX2M+DDz7I6NGjueSSS7xyXiIiUrrSUh1Mn7KVI8m5VKvhz9DR0QQF+1hdloiIiAfLfjPl5uayceNGxowZ415ms9no0qULa9euLXSbjh07smDBAtavX0+7du1ITExk2bJl9O7d22O9IUOGcN1119GlSxeefvrpEqk3JyeHnJwc9+vU1FQAHA4HDoejRI5xtvKPb3UdFZXa17vUvt51PrRvVqaTmc8mcnBfDpWr+HL/I1EEBpXOOZ8P7Wslta/3qY29S+3rXWWpfYtag2XhKzk5GafTSfXqng+7rF69On/++Weh2/Tq1Yvk5GQuvfRSTNMkLy+PwYMHe3QffP/999m0aRMbNmwo0XonT57ME088UWD5ihUrCAoKKtFjna2VK1daXUKFpvb1LrWvd1XU9nXmGWz8JpJjyf74+ju5sP1+1q5LLPU6Kmr7lhVqX+9TG3uX2te7ykL7ZmZmFmm9ctUnY/Xq1UyaNIlZs2bRvn17EhISGDFiBE899RQTJkxg9+7djBgxgpUrVxIQEFCixx4zZgyjRo1yv05NTaVu3bp069aNsLCwEj1WcTkcDlauXEnXrl3x9fW1tJaKSO3rXWpf76rI7ZuX5+L1l3dyLDmdgEAbQ0dHU6d+q1KtoSK3b1mg9vU+tbF3qX29qyy1b36vuDOxLHxVrVoVu93OwYMHPZYfPHiQGjVqFLrNhAkT6N27NwMGDACgefPmZGRkMGjQIMaNG8fGjRtJSkrioosucm/jdDr55ptvmDFjBjk5OdjtZzf42t/fH39//wLLfX19Lf9h5ytLtVREal/vUvt6V0VrX5fL5K1Z2/lzSzp+/jaGPBpNg+gQy+qpaO1b1qh9vU9t7F1qX+8qC+1b1ONbNuGGn58fbdq0YdWqVe5lLpeLVatW0aFDh0K3yczMxGbzLDk/TJmmydVXX82WLVvYvHmz+6tt27bcddddbN68+ayDl4iIlB0ul0n867v4af0xfHwMBo1sSMP/sy54iYiIFJWl3Q5HjRpF3759adu2Le3atWPatGlkZGTQv39/APr06UPt2rWZPHkyALGxsUydOpXWrVu7ux1OmDCB2NhY7HY7oaGhNGvWzOMYwcHBREREeCw/cOAABw4cICEhAYAtW7YQGhpKvXr1qFKlSimdvYiIFJdpmnz47l7Wfn0Yw4D+QxvQpLm1Xb9FRESKytLwdccdd3Do0CEmTpzIgQMHaNWqFZ9//rl7Eo5du3Z53OkaP348hmEwfvx49u7dS2RkJLGxsTzzzDPFOu6cOXM8Js+4/PLLAXjjjTfo16/fuZ+YiIh4xbIPD/DV58efBXn3oPq0uriytQWJiIgUg+UTbgwdOpShQ4cW+t7q1as9Xvv4+BAXF0dcXFyR93/yPgAef/xxHn/88WJUKSIiVvvysySWfbgfgNv61OGSyyMsrkhERKR4LH3IsoiISFGsWZ3MBwv2AHD9f2pyZfdqFlckIiJSfApfIiJSpm1ad5T413cBcPV11bjmpsJnxBURESnrFL5ERKTM+u3nFN6cuQPThE5XRXBzz9oYhmF1WSIiImdF4UtERMqkhL/SeW1aIk6nSZtLwrnznnoKXiIiUq4pfImISJmza3sms59PwJFrcmGrMPrcXx+bTcFLRETKN4UvEREpUw7sy2bmswlkZ7mIviCEASMa4uOjX1ciIlL+6beZiIiUGYcP5TB90lbS0/Ko1zCIwQ81ws9Pv6pERKRi0G80EREpE1KOOnhlcgLHjjqoUTuAIY9GExhkt7osERGREqPwJSIilstIz2PGlK0kH8whItKPYWOiCQn1sbosERGREqXwJSIilsrOcjLzuQT27cmmUmVfho+NoXK4n9VliYiIlDiFLxERsYwj18XclxLZuS2T4BA7Q8dEU7Wav9VliYiIeIXCl4iIWMKZZzJv+nb+/i2NgAAbQx6LpladQKvLEhER8RqFLxERKXUul8k7r+5gy6YUfH0NBj/ciPoNg60uS0RExKsUvkREpFSZpsl/39rNhu+PYrPDgBENiWkSanVZIiIiXqfwJSIipeqThfv49otkDAP63R9Fs9aVrC5JRESkVCh8iYhIqVnxyQFWfHoQgJ731KNNhyoWVyQiIlJ6FL5ERKRUfPPFIT5euA+Am3vVplPnqhZXJCIiUroUvkRExOvWf3+E/765G4BrbqpBl+uqW1yRiIhI6VP4EhERr/pl4zHembMD04QrukVy/X9qWl2SiIiIJRS+RETEa/76LY1507fjckH7y6rwn951MAzD6rJEREQsofAlIiJesT0hgzkvbiPPYdKybSXuGlgfm03BS0REzl8KXyIiUuL27spi1nMJ5Oa4uKBZKP2HNsBuV/ASEZHzm8KXiIiUqKQD2UyfspXMDCcNYoIZNLIhvr76dSMiIqLfhiIiUmKOHs5l+uQE0lLyqF0vkAceaYR/gN3qskRERMoEhS8RESkRaakOpk/ZypHkXKrV9Gfo6GiCgn2sLktERKTMUPgSEZFzlpXpZOazCRzcl0N4hC/DRscQVsnX6rJERETKFIUvERE5J7k5Lma/kMDuHVmEhvkwbEwMVar6WV2WiIhImaPwJSIiZy0vz8Wr0xLZ9lcGgUF2ho6OpnrNAKvLEhERKZMUvkRE5Kw4nSZvztzBH7+k4udv44FHGlGnfpDVZYmIiJRZCl8iIlJsLpdJ/Lxd/LT+GD4+BoNGNqTh/4VYXZaIiEiZpvAlIiLFYpomH767lx++PozNBv2HNqBJ8zCryxIRESnzFL5ERKRYln14gK8+TwLg7kH1aXVxZWsLEhERKScUvkREpMi+/CyJZR/uB+C2vnVof1mExRWJiIiUHwpfIiJSJGtWJ/PBgj0AxN5Wkyu7VbO4IhERkfJF4UtERM5o07qjxL++C4Crr6tG9xtrWFyRiIhI+aPwJSIip/Xbzym8OXMHpgmdrorg5p61MQzD6rJERETKHYUvERE5pYQ/03ltWiJOp0mbS8K58556Cl4iIiJnSeFLREQKtWt7JrNfSMCRa9KsVRh974/CZlPwEhEROVsKXyIiUsCBvVnMfDaB7CwXMU1CuHdEQ+w+Cl4iIiLnQuFLREQ8HD6Uw/TJCaSn5VGvYRD3jWqEn59+XYiIiJwr/TYVERG3lKMOXpmcwLGjDmrUDmDIo9EEBtmtLktERKRCUPgSEREAMtLzmDFlK8kHc4iI9GPYmGhCQn2sLktERKTCUPgSERGys5zMfC6BfXuyqRTuy/CxMVQO97O6LBERkQpF4UtE5DznyHUxd+o2dm7LJDjEzrDR0VSt5m91WSIiIhWOwpeIyHnMmWcyb/p2/v49nYAAG0Mei6ZmnUCryxIREamQFL5ERM5TLpfJ23N3sGVTCr6+BoMfbkT9hsFWlyUiIlJhKXyJiJyHTNNk4Zu7+XHNUWx2GPBgQ2KahFpdloiISIWm8CUich76ZOE+vluVjGFAv/ujaNaqktUliYiIVHgKXyIi55kVnxxgxacHAeh5Tz3adKhicUUiIiLnB4UvEZHzyDdfHOLjhfsAuLlXbTp1rmpxRSIiIucPhS8RkfPE+u8O8983dwNwzU016HJddYsrEhEROb8ofImInAd+/SmVd+buxDThim6RXP+fmlaXJCIict7xsboAERHxrsMH/fnyw124XND+sir8p3cdDMOwuiwREZHzTpm48zVz5kyioqIICAigffv2rF+//rTrT5s2jcaNGxMYGEjdunUZOXIk2dnZha47ZcoUDMPgwQcf9FienZ3NkCFDiIiIICQkhFtvvZWDBw+W1CmJiJQJO7Zlsvn7quTlmbRsW4m7BtbHZlPwEhERsYLl4WvhwoWMGjWKuLg4Nm3aRMuWLenevTtJSUmFrh8fH8/o0aOJi4vjjz/+YN68eSxcuJCxY8cWWHfDhg3MnTuXFi1aFHhv5MiRfPrppyxatIivv/6affv2ccstt5T4+YmIWGXvrkxenboDZ56NxheG0H9oA+x2BS8RERGrWB6+pk6dysCBA+nfvz9NmzZlzpw5BAUFMX/+/ELXX7NmDZ06daJXr15ERUXRrVs3evbsWeBuWXp6OnfddRevvfYa4eHhHu+lpKQwb948pk6dSufOnWnTpg1vvPEGa9as4YcffvDauYqIlJakA9lMn5JAZoaTyhE53DO8Pr6+lv8vX0RE5Lxm6Ziv3NxcNm7cyJgxY9zLbDYbXbp0Ye3atYVu07FjRxYsWMD69etp164diYmJLFu2jN69e3usN2TIEK677jq6dOnC008/7fHexo0bcTgcdOnSxb3sggsuoF69eqxdu5ZLLrmkwHFzcnLIyclxv05NTQXA4XDgcDiKf/IlKP/4VtdRUal9vUvtW/KOHs7llUmJpKXkUbOOPxe03YPN5lQbe4E+v96l9vU+tbF3qX29qyy1b1FrsDR8JScn43Q6qV7dc7rj6tWr8+effxa6Ta9evUhOTubSSy/FNE3y8vIYPHiwR7fD999/n02bNrFhw4ZC93HgwAH8/PyoXLlygeMeOHCg0G0mT57ME088UWD5ihUrCAoKOt1plpqVK1daXUKFpvb1LrVvycjNtrH+q2pkpvkSFOrg/y7ai6+fqfb1MrWvd6l9vU9t7F1qX+8qC+2bmZlZpPXK3WyHq1evZtKkScyaNYv27duTkJDAiBEjeOqpp5gwYQK7d+9mxIgRrFy5koCAgBI77pgxYxg1apT7dWpqKnXr1qVbt26EhYWV2HHOhsPhYOXKlXTt2hVfX19La6mI1L7epfYtOZkZTmY9l0hmWjaVq/gyYlxjQsIMta8X6fPrXWpf71Mbe5fa17vKUvvm94o7E0vDV9WqVbHb7QVmGTx48CA1atQodJsJEybQu3dvBgwYAEDz5s3JyMhg0KBBjBs3jo0bN5KUlMRFF13k3sbpdPLNN98wY8YMcnJyqFGjBrm5uRw7dszj7tfpjuvv74+/v3+B5b6+vpb/sPOVpVoqIrWvd6l9z01ujovXX05kz85sQsN8GD42hmo1AtzdINS+3qX29S61r/epjb1L7etdZaF9i3p8S0df+/n50aZNG1atWuVe5nK5WLVqFR06dCh0m8zMTGw2z7LtdjsApmly9dVXs2XLFjZv3uz+atu2LXfddRebN2/GbrfTpk0bfH19PY77119/sWvXrlMeV0SkrHI4XLw6LZHEvzMIDLIzdHQ01WuW3J1/ERERKRmWdzscNWoUffv2pW3btrRr145p06aRkZFB//79AejTpw+1a9dm8uTJAMTGxjJ16lRat27t7nY4YcIEYmNjsdvthIaG0qxZM49jBAcHExER4V5eqVIl7r33XkaNGkWVKlUICwtj2LBhdOjQodDJNkREyiqn0+TNmTv445dU/PxtPPBoI+rULxvjUEVERMST5eHrjjvu4NChQ0ycOJEDBw7QqlUrPv/8c/ckHLt27fK40zV+/HgMw2D8+PHs3buXyMhIYmNjeeaZZ4p13Jdeegmbzcatt95KTk4O3bt3Z9asWSV6biIi3uRymcTP28XmDcfw8TG4b1RDGsaEWF2WiIiInILl4Qtg6NChDB06tND3Vq9e7fHax8eHuLg44uLiirz/k/cBEBAQwMyZM5k5c2ZxShURKRNM0+TDd/fww9eHsdmg/9AGXNDM2sl/RERE5PT0xE0RkXJo2Yf7+erzQwDcPag+rS6ubG1BIiIickYKXyIi5cyXnx1k2YfHn0l4W986tL8swuKKREREpCgUvkREypE1q5P5YMFeAGJvq8mV3apZXJGIiIgUlcKXiEg5sWndUeJf3wVAl+uq0f3Gwp9LKCIiImWTwpeISDnw288pvDlzB6YJnTpX5aaetTEMw+qyREREpBgUvkREyriEP9N5bVoiTqdJmw7h3Nm/roKXiIhIOaTwJSJShu3ansnsFxJw5Jo0axVG38FR2GwKXiIiIuWRwpeISBl1YG8WM59NIDvLRUyTEO4d0RC7j4KXiIhIeaXwJSJSBh0+lMP0yQmkp+VRv2EQ941qhJ+f/pctIiJSnuk3uYhIGZNy1MErkxM4dtRBjdoBPPBoNIFBdqvLEhERkXOk8CUiUoZkpOcxY8pWkg/mULWaH8PGRBMS6mN1WSIiIlICFL5ERMqI7CwnM59LYN+ebCqF+zJsTAyVw/2sLktERERKiMKXiEgZ4Mh1MXfqNnZuyyQ4xM6w0dFUreZvdVkiIiJSghS+REQs5swzmTd9O3//nk5AgI0hj0VTs06g1WWJiIhICVP4EhGxkMtl8vbcHWzZlIKvr8HghxtRv2Gw1WWJiIiIFyh8iYhYxDRNFr65mx/XHMVmhwEPNiSmSajVZYmIiIiXKHyJiFjk44X7+G5VMoYB/e6PolmrSlaXJCIiIl6k8CUiYoEVnxxg5acHAeh5bz3adKhicUUiIiLibQpfIiKl7JuVh/h44T4Abu5Vm05XVbW4IhERESkNCl8iIqVo/XeHWfjmbgB63FyDLtdVt7giERERKS0KXyIipeTnH4/xztydAFzRLZLrbq1pcUUiIiJSmhS+RERKwZ+/pjJ/+nZcLmh/WRX+07sOhmFYXZaIiIiUIoUvEREv2741g7lTE8nLM2l1cWXuGlgfm03BS0RE5Hyj8CUi4kV7d2Uy87kEcnNcXNAslH5DorDbFbxERETORwpfIiJeknQgm+lTEsjKdNIwJphBIxvi66v/7YqIiJyvdBUgIuIFRw/n8sqkBNJS8qhTP5D7H2mEf4Dd6rJERETEQgpfIiIlLC3FwfTJWzl6OJdqNf0Z+lg0QcE+VpclIiIiFlP4EhEpQZkZecx4NoGD+3MIj/Bj+JgYQiv5Wl2WiIiIlAEKXyIiJSQn28nsF7axZ2cWoWE+DBsTTXiEn9VliYiISBmh8CUiUgIcDhevTUsk8e8MAoPsDBsTTfWaAVaXJSIiImWIwpeIyDlyOk3enLmDP7ak4edv44FHG1G7XpDVZYmIiEgZo/AlInIOXC6T+Nd3snnDMXx8DO4b1ZCGMSFWlyUiIiJlkMKXiMhZMk2TD9/dww/fHMFmg3uGNeCCZmFWlyUiIiJlVLHDV1RUFE8++SS7du3yRj0iIuXGsg/389XnhwC4e1B9WratbG1BIiIiUqYVO3w9+OCDfPjhhzRs2JCuXbvy/vvvk5OT443aRETKrC8/O8iyDw8AcHvfurS/LMLiikRERKSsO6vwtXnzZtavX0+TJk0YNmwYNWvWZOjQoWzatMkbNYqIlClrVifzwYK9AMTeXosrukVaXJGIiIiUB2c95uuiiy7ilVdeYd++fcTFxfH6669z8cUX06pVK+bPn49pmiVZp4hImbBp3VHiXz/e7brLddXofkN1iysSERGR8sLnbDd0OBwsWbKEN954g5UrV3LJJZdw7733smfPHsaOHcsXX3xBfHx8SdYqImKp3zan8ObMHZgmdOpclZt61sYwDKvLEhERkXKi2OFr06ZNvPHGG7z33nvYbDb69OnDSy+9xAUXXOBe5+abb+biiy8u0UJFRKyU8Gc6r72ciNNp0qZDOHf2r6vgJSIiIsVS7PB18cUX07VrV2bPns1NN92Er69vgXUaNGjAnXfeWSIFiohYbdf2TGY/n4Aj16RZqzD6Do7CZlPwEhERkeIpdvhKTEykfv36p10nODiYN95446yLEhEpKw7szWLmswlkZ7uIaRLCvSMaYvdR8BIREZHiK/aEG0lJSaxbt67A8nXr1vHjjz+WSFEiImVBclIOr0xOID0tj/oNgxj8UCP8/PRsehERETk7xb6KGDJkCLt37y6wfO/evQwZMqREihIRsVrKUQfTpySQctRBzToBPPBoNAGBdqvLEhERkXKs2OHr999/56KLLiqwvHXr1vz+++8lUpSIiJXS0/KYPmUryQdzqFrNj6GjowkJPevJYUVERESAswhf/v7+HDx4sMDy/fv34+OjixMRKd+ys5zMej6B/XuyqRTuy7AxMVQO97O6LBEREakAih2+unXrxpgxY0hJSXEvO3bsGGPHjqVr164lWpyISGnKzXUx58Vt7NyWSXCInWGjo6lazd/qskRERKSCKPatqhdeeIHLL7+c+vXr07p1awA2b95M9erVeeedd0q8QBGR0uDMM5k/fTtb/0gnIMDG0NEx1KwTaHVZIiIiUoEUO3zVrl2bX375hXfffZeff/6ZwMBA+vfvT8+ePQt95peISFnncpm8PWcHWzal4OtrMPjhRtRrEGR1WSIiIlLBnNUgreDgYAYNGlTStYiIlDrTNFn45m5+XHsUmx0GPNiQmCahVpclIiIiFdBZz5Dx+++/s2vXLnJzcz2W33DDDedclIhIafl44T6+W5WMYUC/BxrQrFUlq0sSERGRCqrY4SsxMZGbb76ZLVu2YBgGpmkCYBgGAE6ns2QrFBHxkuUfH2Dlp8dnb+15bz3aXBJucUUiIiJSkRV7tsMRI0bQoEEDkpKSCAoK4rfffuObb76hbdu2rF692gslioiUvG9WHuKT/+4D4Ja7atPpqqoWVyQiIiIVXbHvfK1du5Yvv/ySqlWrYrPZsNlsXHrppUyePJnhw4fz008/eaNOEZESs/67wyx8czcAPW6uwdXXVre4IhERETkfFPvOl9PpJDT0+GD0qlWrsm/f8b8c169fn7/++qvYBcycOZOoqCgCAgJo374969evP+3606ZNo3HjxgQGBlK3bl1GjhxJdna2+/3Zs2fTokULwsLCCAsLo0OHDnz22Wce+9i2bRs333wzkZGRhIWFcfvttxf64GgRqXh+/vEY78zdCcCV3SO57taaFlckIiIi54tih69mzZrx888/A9C+fXuee+45vv/+e5588kkaNmxYrH0tXLiQUaNGERcXx6ZNm2jZsiXdu3cnKSmp0PXj4+MZPXo0cXFx/PHHH8ybN4+FCxcyduxY9zp16tRhypQpbNy4kR9//JHOnTtz44038ttvvwGQkZFBt27dMAyDL7/8ku+//57c3FxiY2NxuVzFbQ4RKUf+/DWV+dO343LBJZdX4da767jHq4qIiIh4W7G7HY4fP56MjAwAnnzySa6//nouu+wyIiIiWLhwYbH2NXXqVAYOHEj//v0BmDNnDkuXLmX+/PmMHj26wPpr1qyhU6dO9OrVC4CoqCh69uzJunXr3OvExsZ6bPPMM88we/ZsfvjhBy688EK+//57duzYwU8//URYWBgAb731FuHh4Xz55Zd06dKlWOcgIuXD9q0ZzJ2aSF6eSauLK9NrQH1sNgUvERERKT3FDl/du3d3fx8dHc2ff/7JkSNHCA8PL9ZfkHNzc9m4cSNjxoxxL7PZbHTp0oW1a9cWuk3Hjh1ZsGAB69evp127diQmJrJs2TJ69+5d6PpOp5NFixaRkZFBhw4dAMjJycEwDPz9/d3rBQQEYLPZ+O67704ZvnJycsjJyXG/Tk1NBcDhcOBwOIp83t6Qf3yr66io1L7eVRrtu293FjOfSyQ3x0XjZiHcPag2Llce58PNbn1+vUvt611qX+9TG3uX2te7ylL7FrWGYoUvh8NBYGAgmzdvplmzZu7lVapUKV51QHJyMk6nk+rVPQe6V69enT///LPQbXr16kVycjKXXnoppmmSl5fH4MGDPbodAmzZsoUOHTqQnZ1NSEgIS5YsoWnTpgBccsklBAcH89hjjzFp0iRM02T06NE4nU72799/ynonT57ME088UWD5ihUrCAoKKu7pe8XKlSutLqFCU/t6l7faNyPNhw1fVSM3207liBxq/98eVqz8wyvHKsv0+fUuta93qX29T23sXWpf7yoL7ZuZmVmk9YoVvnx9falXr55lz/JavXo1kyZNYtasWbRv356EhARGjBjBU089xYQJE9zrNW7cmM2bN5OSksLixYvp27cvX3/9NU2bNiUyMpJFixZx//3388orr2Cz2ejZsycXXXQRNtuph8CNGTOGUaNGuV+npqZSt25dunXr5u6+aBWHw8HKlSvp2rUrvr6+ltZSEal9vcub7Xv0cC6vTEokN9tB7XoBDHmsKUHB9hI9Rlmnz693qX29S+3rfWpj71L7eldZat/8XnFnUuxuh+PGjWPs2LG88847Z3XHK1/VqlWx2+0FZhk8ePAgNWrUKHSbCRMm0Lt3bwYMGABA8+bNycjIYNCgQYwbN84dnvz8/IiOjgagTZs2bNiwgZdffpm5c+cC0K1bN7Zt20ZycjI+Pj5UrlyZGjVqnHbCEH9/f4+uivl8fX0t/2HnK0u1VERqX+8q6fZNS3Ew54UdHD3soFpNf4aNjiG00vn789Pn17vUvt6l9vU+tbF3qX29qyy0b1GPX+zwNWPGDBISEqhVqxb169cnODjY4/1NmzYVaT9+fn60adOGVatWcdNNNwHgcrlYtWoVQ4cOLXSbzMzMAnen7Pbjf8U2TfOUx3K5XB7jtfJVrXr8oapffvklSUlJ3HDDDUWqXUTKtsyMPGY8m8DB/TmER/gxfMz5HbxERESkbCh2+MoPSiVh1KhR9O3bl7Zt29KuXTumTZtGRkaGe/bDPn36ULt2bSZPngwcn8lw6tSptG7d2t3tcMKECcTGxrpD2JgxY+jRowf16tUjLS2N+Ph4Vq9ezfLly93HfeONN2jSpAmRkZGsXbuWESNGMHLkSBo3blxi5yYi1sjJdjL7hW3s2ZlFaCUfho+NJjzCz+qyRERERIofvuLi4krs4HfccQeHDh1i4sSJHDhwgFatWvH555+7J+HYtWuXx52u8ePHYxgG48ePZ+/evURGRhIbG8szzzzjXicpKYk+ffqwf/9+KlWqRIsWLVi+fDldu3Z1r/PXX38xZswYjhw5QlRUFOPGjWPkyJEldl4iYg2Hw8Vr0xJJ/DuDwCA7w0ZHU61GgNVliYiIiABnEb5K2tChQ0/ZzXD16tUer318fIiLizttAJw3b94ZjzllyhSmTJlSrDpFpGxzOk3enLmDP7ak4edvY8ij0dSuVzZmIhURERGBswhfNpvttM/zsmomRBE5f7lcJvGv72TzhmP4+BjcN6ohDWKCz7yhiIiISCkqdvhasmSJx2uHw8FPP/3EW2+9VehzsEREvMk0TT5YsIcfvjmCzQb3DGvABc2sffyDiIiISGGKHb5uvPHGAsv+85//cOGFF7Jw4ULuvffeEilMRKQoln6wn9XLDwFw96D6tGxb2dqCRERERE7h1E8VLqZLLrmEVatWldTuRETOaNWyg3y25AAAt/etS/vLIiyuSEREROTUSiR8ZWVl8corr1C7du2S2J2IyBmtWZ3Mh+/uBSD29lpc0S3S4opERERETq/Y3Q7Dw8M9JtwwTZO0tDSCgoJYsGBBiRYnIlKYjT8cJf71XQB0vb463W+obnFFIiIiImdW7PD10ksveYQvm81GZGQk7du3Jzw8vESLExE52W+bU3hr1g5MEy7tXJUb76x12hlYRURERMqKYoevfv36eaEMEZEz2/pHGq9NS8TpNGnbIZw7+tdV8BIREZFyo9hjvt544w0WLVpUYPmiRYt46623SqQoEZGT7dqeyZwXtuFwmDRrHUafwVHYbApeIiIiUn4UO3xNnjyZqlWrFlherVo1Jk2aVCJFiYicaP+eLGZM2Up2touYJiHcO7whdh8FLxERESlfih2+du3aRYMGDQosr1+/Prt27SqRokRE8iUn5TB9SgIZ6U7qNwxi8EON8PMrsadkiIiIiJSaYl/BVKtWjV9++aXA8p9//pmICD1jR0RKTspRB9MnbyXlqIOadQJ44NFoAgLtVpclIiIiclaKHb569uzJ8OHD+eqrr3A6nTidTr788ktGjBjBnXfe6Y0aReQ8lJ6Wx/QpW0lOyqVqNT+GjY4hJLTYcwSJiIiIlBnFvpJ56qmn2LFjB1dffTU+Psc3d7lc9OnTR2O+RKREZGc5mfVcAvv3ZFMp3JdhY2KoFO5rdVkiIiIi56TY4cvPz4+FCxfy9NNPs3nzZgIDA2nevDn169f3Rn0icp7JzXUx58Vt7EzMJDjEzrDR0VSt5m91WSIiIiLn7Kz78MTExBATE1OStYjIec6ZZzLvlUS2/pFOQICNoaNjqFkn0OqyREREREpEscd83XrrrTz77LMFlj/33HPcdtttJVKUiJx/XC6Tt+fs4NefUvH1NRj8cCPqNQiyuiwRERGRElPs8PXNN99w7bXXFljeo0cPvvnmmxIpSkTOL6YJi9/ex49rj2Kzw8AHGxLTJNTqskRERERKVLG7Haanp+Pn51dgua+vL6mpqSVSlIicP0zTZOsvldjx1xEMA/o90IALW1WyuiwRERGRElfsO1/Nmzdn4cKFBZa///77NG3atESKEpHzxxdLD7HjrzAAeg2oR5tLwi2uSERERMQ7in3na8KECdxyyy1s27aNzp07A7Bq1Sri4+NZvHhxiRcoIhXX1ysOsXTxQQBuvLMGHa+sanFFIiIiIt5T7PAVGxvLRx99xKRJk1i8eDGBgYG0bNmSL7/8kipVqnijRhGpgNZ/d5j/vrUbgIZNU7jqmuYWVyQiIiLiXWc11fx1113HddddB0BqairvvfceDz/8MBs3bsTpdJZogSJS8fz84zHembsTgMu7RuBfebfFFYmIiIh4X7HHfOX75ptv6Nu3L7Vq1eLFF1+kc+fO/PDDDyVZm4hUQH/+msr86dtxueCSy6twU8+aGIbVVYmIiIh4X7HufB04cIA333yTefPmkZqayu23305OTg4fffSRJtsQkTNK3JrO3KmJ5OWZtLq4Mr0G1MflyrO6LBEREZFSUeQ7X7GxsTRu3JhffvmFadOmsW/fPqZPn+7N2kSkAtmzM5NZz20jN8dFk+ah9BsShd2uW14iIiJy/ijyna/PPvuM4cOHc//99xMTE+PNmkSkgkk6kM2MKQlkZTpp+H/BDHywIb6+Z93rWURERKRcKvLVz3fffUdaWhpt2rShffv2zJgxg+TkZG/WJiIVwJHkXF6ZlEBaah516gdy/8ON8A+wW12WiIiISKkrcvi65JJLeO2119i/fz/33Xcf77//PrVq1cLlcrFy5UrS0tK8WaeIlENpKQ6mT9nK0cO5VK/pz9DHogkKPqtJVkVERETKvWL3+wkODuaee+7hu+++Y8uWLTz00ENMmTKFatWqccMNN3ijRhEphzIz8pjxbAJJ+3MIj/Bj2JgYQiv5Wl2WiIiIiGXOadBF48aNee6559izZw/vvfdeSdUkIuVcTraT2c9vY8/OLEIr+TB8bDThEX5WlyUiIiJiqRIZ8W6327npppv45JNPSmJ3IlKOORwuXn0pkcStGQQG2Rk2OppqNQKsLktERETEcppuTERKjNNp8ubMHfz5axp+/jaGPBpN7XpBVpclIiIiUiYofIlIiXC5TN59bSebNxzDx8fgvlENaRATbHVZIiIiImWGwpeInDPTNFn8zh7WfXsEmw3uGdaAC5qFWV2WiIiISJmi8CUi52zpB/v5esUhAHrfV5+WbStbW5CIiIhIGaTwJSLn5IulB/lsyQEA7uhXl3aXRlhckYiIiEjZpPAlImft+6+SWRK/F4Abbq/F5V0jLa5IREREpOxS+BKRs7Jx7RHem7cLgK7XV6fbDdUtrkhERESkbFP4EpFi+3VzCm/O3oFpwqWdq3LjnbUwDMPqskRERETKNIUvESmWrX+k8fq0RFxOaNsxnDv611XwEhERESkChS8RKbKdiRnMeWEbDodJs9Zh9LkvCptNwUtERESkKBS+RKRI9u/JYuazCWRnu/i/piHcO7whdh8FLxEREZGiUvgSkTNKTsph+pQEMtKd1G8YxH2jGuHnp/99iIiIiBSHrp5E5LSOHc1l+uStpBx1UKtOAA88Gk1AoN3qskRERETKHYUvETml9LQ8ZkxJIDkpl6rV/Bg6OoaQUB+ryxIREREplxS+RKRQWZlOZj6XwP492VQO92XYmBgqhftaXZaIiIhIuaXwJSIF5Oa6mDt1G7sSMwkOsTNsTDRVq/lbXZaIiIhIuabwJSIenHkm815OZOsf6QQE2hg6OoYatQOtLktERESk3FP4EhE3l8vkrTk7+HVzKr6+Bvc/HE29BkFWlyUiIiJSISh8iQgApmny/hu72bj2KHa7wcAHGxJ9QYjVZYmIiIhUGApfIoJpmnz03l6+/zIZw4B+Q6K4sFUlq8sSERERqVAUvkSE5Z8c5IulSQD0GlCPi9qHW1yRiIiISMVjefiaOXMmUVFRBAQE0L59e9avX3/a9adNm0bjxo0JDAykbt26jBw5kuzsbPf7s2fPpkWLFoSFhREWFkaHDh347LPPPPZx4MABevfuTY0aNQgODuaiiy7igw8+8Mr5iZR1q1ck8el/9wFw69216XhlVYsrEhEREamYLA1fCxcuZNSoUcTFxbFp0yZatmxJ9+7dSUpKKnT9+Ph4Ro8eTVxcHH/88Qfz5s1j4cKFjB071r1OnTp1mDJlChs3buTHH3+kc+fO3Hjjjfz222/udfr06cNff/3FJ598wpYtW7jlllu4/fbb+emnn7x+ziJlybpvD7PorT0A9Li5Bp17VLe4IhEREZGKy9LwNXXqVAYOHEj//v1p2rQpc+bMISgoiPnz5xe6/po1a+jUqRO9evUiKiqKbt260bNnT4+7ZbGxsVx77bXExMTwf//3fzzzzDOEhITwww8/eOxn2LBhtGvXjoYNGzJ+/HgqV67Mxo0bvX7OImXF5g3HWPDqTgCuuiaS626taXFFIiIiIhWbj1UHzs3NZePGjYwZM8a9zGaz0aVLF9auXVvoNh07dmTBggWsX7+edu3akZiYyLJly+jdu3eh6zudThYtWkRGRgYdOnTw2M/ChQu57rrrqFy5Mv/973/Jzs7myiuvPGW9OTk55OTkuF+npqYC4HA4cDgcxTn1Epd/fKvrqKgqYvv+9Vs682fswOWCdpeGE3t7dfLy8iyppSK2b1mi9vUuta93qX29T23sXWpf7ypL7VvUGgzTNE0v11Koffv2Ubt2bdasWeMRjB599FG+/vpr1q1bV+h2r7zyCg8//DCmaZKXl8fgwYOZPXu2xzpbtmyhQ4cOZGdnExISQnx8PNdee637/WPHjnHHHXewYsUKfHx8CAoKYtGiRXTr1u2U9T7++OM88cQTBZbHx8cTFKTnIEn5cSzZjx+/jsTltFGtTiYtLjmMzfLRnyIiIiLlV2ZmJr169SIlJYWwsLBTrmfZna+zsXr1aiZNmsSsWbNo3749CQkJjBgxgqeeeooJEya412vcuDGbN28mJSWFxYsX07dvX77++muaNm0KwIQJEzh27BhffPEFVatW5aOPPuL222/n22+/pXnz5oUee8yYMYwaNcr9OjU1lbp169KtW7fTNnBpcDgcrFy5kq5du+Lr62tpLRVRRWrfvbuymDElEZfTReNmIQwccSE+vtYmr4rUvmWR2te71L7epfb1PrWxd6l9vasstW9+r7gzsSx8Va1aFbvdzsGDBz2WHzx4kBo1ahS6zYQJE+jduzcDBgwAoHnz5mRkZDBo0CDGjRuH7Z8/3/v5+REdHQ1AmzZt2LBhAy+//DJz585l27ZtzJgxg19//ZULL7wQgJYtW/Ltt98yc+ZM5syZU+ix/f398ff3L7Dc19fX8h92vrJUS0VU3tv34P5s5rywg6xMFw3/L5j7RjbCP8BudVlu5b19yzq1r3epfb1L7et9amPvUvt6V1lo36Ie37I/efv5+dGmTRtWrVrlXuZyuVi1apVHN8QTZWZmugNWPrv9+MXj6XpPulwu93itzMxMgEL343K5in8iIuXAkeRcpk/eSlpqHnXqB3L/w2UreImIiIicDyztdjhq1Cj69u1L27ZtadeuHdOmTSMjI4P+/fsDx6eEr127NpMnTwaOz2Q4depUWrdu7e52OGHCBGJjY90hbMyYMfTo0YN69eqRlpZGfHw8q1evZvny5QBccMEFREdHc9999/HCCy8QERHBRx99xMqVK/nf//5nTUOIeFFqioPpU7Zy9LCD6jX9GfpYNEHB5arHsYiIiEiFYOkV2B133MGhQ4eYOHEiBw4coFWrVnz++edUr378WUO7du3yuEM1fvx4DMNg/Pjx7N27l8jISGJjY3nmmWfc6yQlJdGnTx/2799PpUqVaNGiBcuXL6dr167A8VuCy5YtY/To0cTGxpKenk50dDRvvfWWx6QcIhVBZkYeM6YkkLQ/h/AIP4aNiSG0kro9iIiIiFjB8j9/Dx06lKFDhxb63urVqz1e+/j4EBcXR1xc3Cn3N2/evDMeMyYmhg8++KBYdYqUNznZTmY/v429u7IIreTD8LHRhEf4WV2WiIiIyHlLE0yLVEAOh4tXX0okcWsGgUF2ho2OplqNAKvLEhERETmvKXyJVDBOp8kbM7bz569p+PnbGPJoNLXr6Vl0IiIiIlZT+BKpQFwuk3df28nPP6bg42Nw36iGNIgJtrosEREREUHhS6TCME2Txe/sYd23R7DZ4J5hDbigmbUPABcRERGRfyl8iVQQSz/Yz9crDgHQ+776tGxb2dqCRERERMSDwpdIBfDF0oN8tuQAAHf0q0u7SyMsrkhERERETqbwJVLOff9lMkvi9wJww+21uLxrpMUViYiIiEhhFL5EyrGNa4/w3vxdAHSNrU73G2tYXJGIiIiInIrCl0g59evmFN6cvQPThEuvrsqNd9SyuiQREREROQ2FL5FyaOsfabw+LRGXE9p2DOeOfnUxDMPqskRERETkNBS+RMqZnYkZzHlhGw6HSfOLKtHnvihsNgUvERERkbJO4UukHNm/J4uZzyaQne3i/5qGcO+wBth9FLxEREREygOFL5FyIjkph+mTE8hId1K/URD3jWqEr5/+CYuIiIiUF7pyEykHjh3NZfrkraQcc1CrTgBDHo0mINBudVkiIiIiUgwKXyJlXHpaHtMnJ5CclEvV6v4MHR1DcIiP1WWJiIiISDEpfImUYVmZTmY+l8CBvdlUDvdl+JhoKoX7Wl2WiIiIiJwFhS+RMio318XcqdvYlZhJSKgPw8ZEExHpb3VZIiIiInKWFL5EyqC8PBfzXk5k6x/pBATaGPJYNDVqB1pdloiIiIicA4UvkTLG5TJ5e/ZOft2ciq+fwf0PR1OvQZDVZYmIiIjIOVL4EilDTNPk/fm72PjDUex2g4EPNiT6ghCryxIRERGREqDwJVJGmKbJkvf28v1XhzEM6DckigtbVrK6LBEREREpIQpfImXE8o8PsGppEgC9BtTjovbhFlckIiIiIiVJ4UukDFi9IolPF+0H4Na7a9PxyqoWVyQiIiIiJU3hS8Ri6749zKK39gBw7S016NyjusUViYiIiIg3KHyJWGjzhmO8M3cnAFddE8m1t9S0uCIRERER8RaFLxGL/LEllTdmbMc04ZIrIrjlrjoYhmF1WSIiIiLiJQpfIhZI3JrOqy8lkpdn0uriyvS6tx42m4KXiIiISEWm8CVSyvbszGTWc9vIzXHRpEUY/YZEYbcreImIiIhUdApfIqXo4P5sZkxJICvTScP/C2bgiAb4+uqfoYiIiMj5QFd9IqXkSHIu0ydvJS01j7pRgTzwSDT+AXaryxIRERGRUqLwJVIKUlMcTJ+ylaOHHVSv6c+QR6MJDFLwEhERETmfKHyJeFlmRh4zpiSQtD+HKlX9GDYmhtBKvlaXJSIiIiKlTOFLxItysp3Men4be3dlEVrJh2FjogmP8LO6LBERERGxgMKXiJc4HC5efSmR7VszCAq2M2x0DNVqBFhdloiIiIhYROFLxAucTpM3Zmznz1/T8PO38cAj0dSuF2h1WSIiIiJiIYUvkRLmcpm8+9pOfv4xBR9fg8EPNaJBTLDVZYmIiIiIxRS+REqQaZosfmcP6749gs0G9wxtQOMLQ60uS0RERETKAIUvkRL0v8X7+XrFIQwDeg+OomXbylaXJCIiIiJlhMKXSAn5YulBPv/oAAC3961Lu05VLK5IRERERMoShS+REvD9l8ksid8LwI131OLyrpEWVyQiIiIiZY3Cl8g52rj2CO/N3wVA19jqdLuhhsUViYiIiEhZpPAlcg5+/SmFN2fvwDTh0qurcuMdtawuSURERETKKIUvkbO09Y80Xn85EZcT2nYM545+dTEMw+qyRERERKSMUvgSOQs7EzOY88I2HA6T5hdVos99UdhsCl4iIiIicmoKXyLFtG9PFjOfTSA728X/NQ3h3mENsPsoeImIiIjI6Sl8iRRDclIOMyYnkJHupH6jIO4b1QhfP/0zEhEREZEz87G6AJHyIuWog1cmJZJyzEGtOgEMeTSagEC71WWJiIiISDmhP9mLFEFujo3Zz2/n8KFcqlb3Z+joGIJD9LcLERERESk6XT2KnEF2lpNN30SSejSHyuG+DB8TTaVwX6vLEhEREZFyRne+RE4jN9fFa9N2knrUj+BQO8PGRBMR6W91WSIiIiJSDil8iZxCXp6L119OZNtfGfj4uhj8UANq1A60uiwRERERKacUvkQK4XKZvD17J79tTsXXz6D1pYeoG6XgJSIiIiJnT+FL5CSmafL+/F1s/OEodrvBPUPrEx6Za3VZIiIiIlLOlYnwNXPmTKKioggICKB9+/asX7/+tOtPmzaNxo0bExgYSN26dRk5ciTZ2dnu92fPnk2LFi0ICwsjLCyMDh068Nlnn7nf37FjB4ZhFPq1aNEir52nlH2mabLkvb18/9VhDAP6DYmiSYtQq8sSERERkQrA8vC1cOFCRo0aRVxcHJs2baJly5Z0796dpKSkQtePj49n9OjRxMXF8ccffzBv3jwWLlzI2LFj3evUqVOHKVOmsHHjRn788Uc6d+7MjTfeyG+//QZA3bp12b9/v8fXE088QUhICD169CiV85ayafnHB1i19Phnr9eAelzUPtziikRERESkorB8qvmpU6cycOBA+vfvD8CcOXNYunQp8+fPZ/To0QXWX7NmDZ06daJXr14AREVF0bNnT9atW+deJzY21mObZ555htmzZ/PDDz9w4YUXYrfbqVGjhsc6S5Ys4fbbbyckJKSkT1HKidXLk/h00X4Abr27Dh2vrGpxRSIiIiJSkVgavnJzc9m4cSNjxoxxL7PZbHTp0oW1a9cWuk3Hjh1ZsGAB69evp127diQmJrJs2TJ69+5d6PpOp5NFixaRkZFBhw4dCl1n48aNbN68mZkzZ56y1pycHHJyctyvU1NTAXA4HDgcjjOeqzflH9/qOsqz9d8fZdHbewDofmM1LusSXqBd1b7eofb1LrWvd6l9vUvt631qY+9S+3pXWWrfotZgmKZpermWU9q3bx+1a9dmzZo1HsHo0Ucf5euvv/a4m3WiV155hYcffhjTNMnLy2Pw4MHMnj3bY50tW7bQoUMHsrOzCQkJIT4+nmuvvbbQ/T3wwAOsXr2a33///ZS1Pv744zzxxBMFlsfHxxMUFFSU05Uy6uCeQH5eGwGmQb2YNBq3OoZhWF2ViIiIiJQXmZmZ9OrVi5SUFMLCwk65nuXdDotr9erVTJo0iVmzZtG+fXsSEhIYMWIETz31FBMmTHCv17hxYzZv3kxKSgqLFy+mb9++fP311zRt2tRjf1lZWcTHx3tsW5gxY8YwatQo9+vU1FTq1q1Lt27dTtvApcHhcLBy5Uq6du2Kr6+vpbWUN3/9msaqD3eCadLusnDu7N8Mm80zeal9vUvt611qX+9S+3qX2tf71Mbepfb1rrLUvvm94s7E0vBVtWpV7HY7Bw8e9Fh+8ODBAmOy8k2YMIHevXszYMAAAJo3b05GRgaDBg1i3Lhx2GzH5xDx8/MjOjoagDZt2rBhwwZefvll5s6d67G/xYsXk5mZSZ8+fU5bq7+/P/7+/gWW+/r6Wv7DzleWaikPEv9OZ970XTjzTFq3q0zvQVEFgteJ1L7epfb1LrWvd6l9vUvt631qY+9S+3pXWWjfoh7f0tkO/fz8aNOmDatWrXIvc7lcrFq16pTjszIzM90BK5/dbgeOTxN+Ki6Xy2PMVr558+Zxww03EBkZeTanIOXUnp2ZzHp+G7k5Lpq0CKPvA6cPXiIiIiIi58ryboejRo2ib9++tG3blnbt2jFt2jQyMjLcsx/26dOH2rVrM3nyZOD4TIZTp06ldevW7m6HEyZMIDY21h3CxowZQ48ePahXrx5paWnEx8ezevVqli9f7nHshIQEvvnmG5YtW1a6Jy2WOrg/mxlTEsjKdNKocTCDHmyIr6/lT10QERERkQrO8vB1xx13cOjQISZOnMiBAwdo1aoVn3/+OdWrVwdg165dHne6xo8fj2EYjB8/nr179xIZGUlsbCzPPPOMe52kpCT69OnD/v37qVSpEi1atGD58uV07drV49jz58+nTp06dOvWrXROVix3JDmX6ZO3kpaaR92oQO5/OBo/fwUvEREREfE+y8MXwNChQxk6dGih761evdrjtY+PD3FxccTFxZ1yf/PmzSvScSdNmsSkSZOKXKeUb6kpDqZP3srRww6q1/JnyGPRBAbZrS5LRERERM4T+pO/nBcyM/KYMSWBpAM5VKnqx7DRMYSGaeCriIiIiJQehS+p8LKzncx6fht7d2URWsmHYWOiCY/ws7osERERETnPKHxJheZwuHjtpUS2b80gKNjOsNExVKsRYHVZIiIiInIeUviSCsvpNHljxnb+/DUNP38bDzwaTe16gVaXJSIiIiLnKYUvqZBcLpN3X9vJzz+m4ONrMPihRjSIDra6LBERERE5jyl8SYVjmiaL39nDum+PYLPBvcMa0PjCUKvLEhEREZHznMKXVDj/W7yfr1ccwjCg9+AoWrSpbHVJIiIiIiIKX1KxfLH0IJ9/dACA2/vVpV2nKhZXJCIiIiJynMKXVBjff5nMkvi9ANx4Ry0u7xJpcUUiIiIiIv9S+JIK4ce1R3hv/i4AusVWp9sNNSyuSERERETEk8KXlHu//pTCW7N3YJpwWZeq3HBHLatLEhEREREpQOFLyrWtf6Tx+suJuJzQtmM4t/eti2EYVpclIiIiIlKAwpeUWzu3ZTD7hW04HCbNL6pEn/uisNkUvERERESkbFL4knJp354sZj6XQE62i/9rGsK9wxpg91HwEhEREZGyS+FLyp3kpBxmTE4gI91JVKMg7hvVCF8/fZRFREREpGzTFauUK8eO5PLKpK2kHHNQq04ADzwaTUCg3eqyRERERETOSOFLyo30tDymT0ng8KFcqlb3Z+joGIJDfKwuS0RERESkSBS+pFzIynQy89kEDuzNpnK4L8PHRFMp3NfqskREREREikzhS8q83FwXc17cxq7tmYSE+jBsbAwRkf5WlyUiIiIiUiwKX1Km5eW5eP3lRBL+TCcg0MaQx6KpUSvA6rJERERERIpN4UvKLJfL5O3ZO/ltcyq+fgb3PxJNvQZBVpclIiIiInJWFL6kTDJNk/fm7WLjD0ex2w0GPtiQ6MYhVpclIiIiInLWFL6kzDFNkyXxe1mz+jCGAf2GRHFhy0pWlyUiIiIick4UvqTM+fyjA6xalgRArwH1uKh9uMUViYiIiIicO4UvKVNWL0/if4v3A3Dr3XXoeGVViysSERERESkZCl9SZvzwzWEWvb0HgGtvqUnnHtUsrkhEREREpOQofEmZsHnDURa8uhOAq66pxrW31LC4IhERERGRkqXwJZb7Y0sqb8zYgWlChysiuPXu2hiGYXVZIiIiIiIlSuFLLJX4dzqvvpRIXp5J63aV6TWgnoKXiIiIiFRICl9imd07Mpn1/DZyc1w0bRFGvyFR2GwKXiIiIiJSMSl8iSUO7s9mxrMJZGU6adQ4mIEPNsTHRx9HEREREam4dLUrpe5Ici7TJ28lPTWPulGB3P9wNH7++iiKiIiISMWmK14pVakpDqZP3srRww6q1/JnyGPRBAbZrS5LRERERMTrFL6k1GRm5DFjSgJJB3KoUtWPYaNjCA3ztbosEREREZFSofAlpSI728ms57exd1cWYZV9GDYmmvAIP6vLEhEREREpNQpf4nWOXBevvZTI9q0ZBAXbGfpYDNVqBFhdloiIiIhIqVL4Eq9yOk3emLmdP39Nw8/fxgOPRlO7XqDVZYmIiIiIlDqFL/Eal8tkwas7+fnHFHx8De5/qBENooOtLktERERExBIKX+IVpmmy+J09rP/uCDYb3DusAf93YajVZYmIiIiIWEbhS7zif4v28/WKQxgG9BkcRYs2la0uSURERETEUgpfUuJW/u8gn398AIA7+tXl4k5VLK5IRERERMR6Cl9Sor77MpmP3tsLwI131uKyLpEWVyQiIiIiUjYofEmJ+XHtEd6fvwuAbrHV6RZbw+KKRERERETKDoUvKRG//pTCW7N3YJpwWZeq3HBHLatLEhEREREpUxS+5Jz9/Xsar7+ciMsJF3cK5/a+dTEMw+qyRERERETKFIUvOSc7t2Uw58VtOBwmzS+qRO9BUdhsCl4iIiIiIidT+JKztm9PFjOfSyAn28X/XRjKvcMaYPdR8BIRERERKYyP1QVI+ZSclMOMyQlkpDuJahTEfSMb4uunLC8iInI+czqdOBwOq8soMxwOBz4+PmRnZ+N0Oq0up8Ipzfb19fXFbref834UvqTYjh3J5ZVJW0k55qBW3QAeeDSagMBz/zCKiIhI+WSaJgcOHODYsWNWl1KmmKZJjRo12L17t8bDe0Fpt2/lypWpUaPGOR1L4UuKJT0tj+mTEzh8KJeq1f0ZOjqG4BB9jERERM5n+cGrWrVqBAUFKWj8w+VykZ6eTkhICDabegiVtNJqX9M0yczMJCkpCYCaNWue9b501SxFlpXpZOazCRzYl03lKr4MHxNNpcq+VpclIiIiFnI6ne7gFRERYXU5ZYrL5SI3N5eAgACFLy8ozfYNDAwEICkpiWrVqp11F0R9CqRIcnNczHlxG7u2ZxIS6sOwMTFERPpbXZaIiIhYLH+MV1BQkMWViHhX/mf8XMY1KnzJGeXluXjt5UQS/kwnINDGkMeiqVErwOqyREREpAxRV0Op6EriM255+Jo5cyZRUVEEBATQvn171q9ff9r1p02bRuPGjQkMDKRu3bqMHDmS7Oxs9/uzZ8+mRYsWhIWFERYWRocOHfjss88K7Gft2rV07tyZ4OBgwsLCuPzyy8nKyirx8yvvXC6Tt2bt4PefU/H1M7j/kWjqNdBftkREREREisvS8LVw4UJGjRpFXFwcmzZtomXLlnTv3t09mO1k8fHxjB49mri4OP744w/mzZvHwoULGTt2rHudOnXqMGXKFDZu3MiPP/5I586dufHGG/ntt9/c66xdu5ZrrrmGbt26sX79ejZs2MDQoUPVF/ckpmny3rxdbFp3DLvdYNDIRkQ3DrG6LBEREZEyKyoqimnTphV5/dWrV2MYhmaKPE9YOuHG1KlTGThwIP379wdgzpw5LF26lPnz5zN69OgC669Zs4ZOnTrRq1cv4PiHu2fPnqxbt869TmxsrMc2zzzzDLNnz+aHH37gwgsvBGDkyJEMHz7c4xiNGzc+ba05OTnk5OS4X6empgLH+3xa/TyL/OOXZB2mafLJwgOsWX0Yw4Deg+sS0yTQ8nO1gjfaV/6l9vUuta93qX29S+3rfSXRxg6HA9M0cblcuFyukirN6840YcLEiROJi4sr9n7XrVtHcHAwLpcL0zQB3O1TmEsuuYS9e/cSGhpaau3XtGlTtm/fzvbt26lRo0apHNMbitK+JSn/Z+pwOAp8for6b8gw86suZbm5uQQFBbF48WJuuukm9/K+ffty7NgxPv744wLbxMfH88ADD7BixQratWtHYmIi1113Hb179/a4+5XP6XSyaNEi+vbty08//UTTpk1JSkqievXqvPLKK7z33nts27aNCy64gGeeeYZLL730lPU+/vjjPPHEE4XWVBEHmCb+HkbCr5UAuPDiI9RukGFxRSIiIlIW+fj4UKNGDerWrYufn5/V5RTZwYMH3d8vWbKESZMmsWHDBvey4OBgQkKO9/gxTROn04mPT/mfKHzt2rUMGjSISy65hAsvvJAHH3zQ0nocDge+vuVj9uzc3Fx2797NgQMHyMvL83gvMzOTXr16kZKSQlhY2Cn3YdknKDk5GafTSfXq1T2WV69enT///LPQbXr16kVycjKXXnoppmmSl5fH4MGDCwSvLVu20KFDB7KzswkJCWHJkiU0bdoUgMTEROB4mHrhhRdo1aoVb7/9NldffTW//vorMTExhR57zJgxjBo1yv06NTWVunXr0q1bt9M2cGlwOBysXLmSrl27lsiH9+uVyST8uh+Am3vV5Ipuzc95n+VZSbeveFL7epfa17vUvt6l9vW+kmjj7Oxsdu/eTUhICAEBxyfkMk2T3Bxr7oL5+duKNDHCiddv1apVw2azua8DV69ezdVXX83//vc/Jk6cyJYtW/j888+pW7cuDz30EOvWrSMjI4MmTZrwzDPP0KVLF/e+GjZsyIgRIxgxYgSmaeLj48PcuXNZtmwZK1asoHbt2jz//PPccMMNHsc6fPgwlStX5s0332TUqFG89957jBo1it27d9OpUyfmz5/vfr5UXl4eDz30EO+88w52u517772XAwcOkJKSwpIlS0573gsXLuSuu+7i8ssvZ+TIkUycONHj/T179vDoo4+yYsUKcnJyaNKkCdOnT6d9+/YAfPrppzz99NNs2bKFkJAQLr30Uj788EPg+N3EDz74wOPGSpUqVZg6dSr9+vVjx44dNGrUiPj4eObMmcO6deuYNWsWsbGxDBs2jG+//ZajR4/SqFEjRo8eTc+ePd37cblcvPjii7z22mvs3r2b6tWrM3DgQIYNG8bNN99M06ZNmT59unv9Q4cOUbduXZYuXcrVV199xs9DUWRnZxMYGMjll1/u/qzny+8VdyblKr6vXr2aSZMmMWvWLNq3b09CQgIjRozgqaeeYsKECe71GjduzObNm0lJSWHx4sX07duXr7/+mqZNm7pvSd53333u7o6tW7dm1apVzJ8/n8mTJxd6bH9/f/z9C06t7uvrW2Z+IZRELT98c5gl7x4PXtfeUpMu1539Q+QqmrL0s66I1L7epfb1LrWvd6l9ve9c2tjpdGIYBjabzT1+PifbycMDt5RkiUU2dV5L/AOKN44/v+6T/zt27FheeOEFGjZsSHh4OLt37+a6665j0qRJ+Pv78/bbb3PjjTfy119/Ua9ePff+8tsj/7rzqaee4rnnnuOFF15g+vTp9O7dm507d1KlShWPY+Z/ZWZmMnXqVN555x1sNht33303jz76KO+++y4Azz//PPHx8bzxxhs0adKEl19+mY8//pirrrrqtHMYpKWlsXjxYtatW8cFF1xASkoK33//PZdddhkA6enpXHXVVdSuXZtPPvmEGjVqsGnTJnd9S5cu5dZbb2XcuHG8/fbb5ObmsmzZMo9jnvg5OHnZie364osv0rp1awICAsjNzaVt27aMHj2asLAwli5dSt++fYmJiaFdu3bA8Rshr732Gi+99BKXXnop+/fv5/fffwfg3nvvZfjw4UydOtV9vR4fH0/t2rXp0qVLic3EabMdD/aF/Xsp6r8fy8JX1apVsdvtHrd84fgt4FP1PZ0wYQK9e/dmwIABADRv3pyMjAwGDRrEuHHj3D9QPz8/oqOjAWjTpg0bNmzg5ZdfZu7cue6/GOTfCcvXpEkTdu3aVaLnWN5s3nCUBa/uBKBzj2pce0v57QMsIiIicq6efPJJunbt6n5dpUoVWrZs6X791FNPsWTJEj755BOGDh16yv307dvXfRdn0qRJvPLKK6xfv55rrrmm0PUdDgdz5syhUaNGAAwdOpQnn3zS/f706dMZM2YMN998MwAzZsxg2bJlZzyf999/n5iYGPc8CHfeeSfz5s1zh6/4+HgOHTrEhg0bqFKlCoD7mhqOz6Vw5513egzFObE9iurBBx/klltu8Vj28MMPu78fNmwYy5cv57///S/t2rUjLS2Nl19+mRkzZtC3b18AGjVqRMeOHUlNTeWWW25h+PDhfPzxx9x+++0AvPnmm/Tr16/MPQLBsvDl5+dHmzZtWLVqlfvWpMvlYtWqVaf88GZmZhZI0vmD3U43dM3lcrkny4iKiqJWrVr89ddfHuv8/fff9OjR42xPp9z745dU5k/fgWlChysiuOWu2mXuwyoiIiLlg5+/janzin9RXlLHLilt27b1eJ2ens7jjz/O0qVL2b9/P3l5eWRlZZ3xD/jNm/87hCP/MUenmt0bjj/MNz94AdSsWdO9fkpKCgcPHnTfEYLj18Nt2rQ546QT8+fP5+6773a/vvvuu7niiiuYPn06oaGhbN68mdatW7uD18k2b97MwIEDT3uMoji5XZ1OJ5MmTeK///0ve/fuJTc3l5ycHPe8Cn/88Qc5OTmn7D4YEBBA7969mT9/PrfffjubNm3i119/5ZNPPjnnWkuapd0OR40aRd++fWnbti3t2rVj2rRpZGRkuLsD9unTh9q1a7u7AsbGxjJ16lRat27t7nY4YcIEYmNj3SFszJgx9OjRg3r16pGWlkZ8fDyrV69m+fLlwPHbwI888ghxcXG0bNmSVq1a8dZbb/Hnn3+yePFiaxrCYtv+SufVaYk4nSat21em14B6Cl4iIiJy1gzDwD/g9LMJlgfBwcEerx9++GFWrlzJCy+8QHR0NIGBgfznP/8hNzf3tPs5uUuaYRinDUqFrX+uc+T9/vvv/PDDD6xfv57HHnvMvdzpdPL+++8zcOBAAgMDT7uPM71fWJ2FzQJ4crs+//zzvPzyy0ybNo3mzZsTHBzMgw8+6G7XMx0XYMCAAbRq1Yo9e/bwxhtv0LlzZ+rXr3/G7UqbpeHrjjvu4NChQ0ycOJEDBw7QqlUrPv/8c/ckHLt27fK40zV+/HgMw2D8+PHs3buXyMhIYmNjeeaZZ9zrJCUl0adPH/bv30+lSpVo0aIFy5cv97hl/OCDD5Kdnc3IkSM5cuQILVu2ZOXKlR5/YThf7N6RyewXtpGb46JpizD6PRCFzabgJSIiInKy77//nn79+rm7+6Wnp7Njx45SraFSpUpUr16dDRs2cPnllwPHA9SmTZto1arVKbebN28el19+OTNnzvRY/sYbbzBv3jwGDhxIixYteP311zly5Eihd79atGjBqlWr3DdKThYZGcn+/fvdr7du3UpmZuYZz+n777/nxhtvdN+Vc7lc/P333+5hQjExMQQGBrJq1Sr38KOTNW/enLZt2/Laa68RHx/PjBkzznhcK1g+4cbQoUNP2c1w9erVHq99fHyIi4s77TMX5s2bV6Tjjh49utBniZ1PDu7LZsazCWRlOmnUOJiBDzbEx0cPmhYREREpTExMDB9++CGxsbEYhsGECRMsebbZsGHDmDx5MtHR0VxwwQVMnz6do0ePnrLnksPh4J133uHJJ5+kWbNmHu8NGDCAqVOn8ttvv9GzZ08mTZrETTfdxOTJk6lZsyY//fQTtWrVokOHDsTFxXH11VfTqFEj7rzzTvLy8li2bJn7Tlrnzp2ZMWMGHTp0wOl08thjjxVpIoqYmBgWL17MmjVrCA8PZ+rUqRw8eNAdvgICAnjsscd49NFH8fPzo1OnThw6dIgtW7Zw2223eZzL0KFDCQ4OdgfkskZX2uepI8m5TJ+ylfTUPOpGBXL/w9El2kdaREREpKKZOnUq4eHhdOzYkdjYWLp3785FF11U6nU89thj9OzZkz59+tChQwdCQkLo3r17genP833yySccPny40EDSpEkTmjRpwrx58/Dz82PFihVUq1aNa6+9lubNmzNlyhT38J4rr7ySRYsW8cknn9CqVSs6d+7M+vXr3ft68cUXqVu3Lpdddhm9evXi4YcfLtLzcMePH89FF11E9+7dufLKK6lRo4bHdPVwfOK9hx56iIkTJ9KkSRN3D7oT9ezZEx8fH3r27HnKtrCaZQ9ZLu9SU1OpVKnSGR+kVhocDgfLli3j2muvLdJfF1JTHLz05N8kHcihei1/Rk74P0LDNIXvqRS3faV41L7epfb1LrWvd6l9va8k2jg7O5vt27fToEGDMnvBaxWXy0VqaiphYWGnnQK+JI7TpEkTbr/9dp566imvHaesObl9858jtmHDBq+E4tN91ouaDSzvdiilKzMjjxlTEkg6kEOVqn4MGx2j4CUiIiJSjuzcuZMVK1ZwxRVXkJOTw4wZM9i+fTu9evWyujRLOBwOjh49yvjx47nkkkssuRtZVOpndh7JznYy67lt7N2VRVhlH4aNiSY8ws/qskRERESkGGw2G2+++SYXX3wxnTp1YsuWLXzxxRc0adLE6tIs8f3331OzZk02bNjAnDlzrC7ntHTn6zzhyHXx6tREtidkEBRsZ+hjMVSroa4BIiIiIuVN3bp1+f77760uo8y48sorz3kq/tKiO1/nAafT5I2Z2/nrtzT8A2wMeSya2vXO/LwEEREREREpOQpfFZzLZbLg1Z38/GMKPr4Gg0c1IqpR8Jk3FBERERGREqXwVYGZpsmit/ew/rsj2GwwYHgD/u/CUKvLEhERERE5Lyl8VWD/W7Sfb1YewjCgz+Aoml9U2eqSRERERETOWwpfFdTK/x3k848PwP+3d+9RVVR7HMC/w+vwkJeCcFADEQQlwCuKoZkYKKCRtCzBkCBfPcSVlaVlvtJbVJRpGd26PDITwgwr8xEqkCFmqSgWEXIRzUdkpjwUBM6+f7A8NQIKyDmHx/ez1lly9uyZ2fPrx3R+zMw+AMJjBmDkmN46HhERERERUc/G4qsb+m7vBWxNPQMAmBLhgLGBtjoeERERERERsfjqZn7cfxFpSacAABPvt8PEUHsdj4iIiIio+/L398eCBQvU752cnPD222/fdB1JkrB169bb3ndHbYe0h8VXN/JTfgU+ev8khADuCbTB/dMcdD0kIiIiok4pNDQUwcHBzS7bt28fJEnCsWPH2rzdH374AXPnzr3d4cmsWLECw4YNa9J+7tw5hISEdOi+WnL16lX07t0bNjY2qK2t1co+uyMWX93ExXIFUtafgqoBGDmmNx6KHgBJknQ9LCIiIqJOadasWcjMzMRvv/3WZFlycjJGjBgBLy+vNm/X1tYWpqamHTHEW7K3t4dCodDKvrZs2QIPDw+4u7vr/GqbEAL19fU6HUN7sfjqBk797wqOfGeDujoBLx9LRM11hJ4eCy8iIiLSDSEE6kSdTl5CiFaN8b777oOtrS1SUlJk7VVVVdi8eTNmzZqFP//8E9OnT0e/fv1gamoKT09PpKam3nS7N952WFJSAn9/fxgbG2Po0KHIzMxsss6iRYswePBgmJqawtnZGUuXLkVdXR0AICUlBStXrsTRo0chSRIkSVKP+cbbDgsKCnDvvffCxMQEffr0wdy5c1FVVaVeHhMTg7CwMMTHx0OpVKJPnz6YN2+eel83k5iYiBkzZmDGjBlITExssvynn37CfffdBwsLC5ibm2Ps2LEoKSlRL09KSoKHhwcUCgWUSiViY2MBACdPnoQkScjPz1f3vXTpEiRJQnZ2NgAgOzsbkiRhx44d8PHxgUKhwHfffYeSkhI8/PDDUCqV6NWrF0aOHIndu3fLxlVbW4tFixZhwIABUCgUcHFxQWJiIoQQcHFxQXx8vKx/fn4+JEnCiRMnbhmT9jDQyFZJay5euIb33zyJhno9uA4xw8zYgdA3YOFFREREulOPerx36T2d7PtJqydhCMNb9jMwMMAjjzyClJQULFmyRH3H0ObNm9HQ0IDp06ejqqoKPj4+WLRoESwsLPD1118jKioKgwYNgq+v7y33oVKpEBUVBQcHB3z//fe4fPmy7Pmw68zNzZGSkgIHBwcUFBRgzpw5MDc3x/PPP4/w8HAcP34cO3fuVBcWlpaWTbZRXV2NoKAg+Pn54YcffkB5eTlmz56N2NhYWYGZlZUFpVKJrKwsnDhxAuHh4Rg2bBjmzJnT4nGUlJQgLy8Pn3/+OYQQePrpp1FWVgZHR0cAwJkzZ3DPPffA398fe/fuhYWFBXJzc9VXpxISEvDMM88gLi4OISEhuHz5MnJzc28ZvxstXrwY8fHxcHZ2hrW1NcrKyjBhwgTExcXBxMQEGzZsQGhoKIqKinDHHXcAAB555BHk5eVh3bp18Pb2RmlpKS5cuABJkjBz5kwkJydj4cKF6n0kJyfjnnvugYuLS5vH1xosvro4q96GGDHaCkcPncXsp4bC0IgXM4mIiIhaY+bMmXjjjTeQk5MDf39/AI0fvqdOnQpLS0tYWlrKPpjPnz8fu3btQnp6equKr927d6O4uBjffPMN+vfvDwB45ZVXmjyn9dJLL6l/dnJywsKFC5GWlobnn38eJiYm6NWrFwwMDGBv3/JEaps2bUJNTQ02bNgAMzMzAMC7776L0NBQvPbaa7CzswMAWFtb491334W+vj7c3d0xefJk7Nmz56bFV1JSEkJCQmBtbQ0ACAoKQnJyMlasWAEAWL9+PSwtLZGWlgZDw8bCd/Dgwer1V69ejWeffRZPPfWUum3kyJG3jN+NXn75ZUyYMEH93srKCgMHDoSFhQX09PSwatUqZGRk4Msvv0RsbCx+/fVXpKenIzMzE4GBgQAAZ2dn9foxMTFYtmwZDh48CF9fX9TV1WHTpk1NroZ1JBZfXZyenoQHHlbC0PwoFMb6uh4OEREREQxggCetntTZvlvL3d0do0ePRlJSEvz9/XHixAns27cPL7/8MgCgoaEBr7zyCtLT03HmzBlcu3YNtbW1rX6m65dffkG/fv3g4PD3JGh+fn5N+n366adYt24dSkpKUFVVhfr6elhYWLT6OACgsLAQ3t7e6sILAMaMGQOVSoWioiJ18eXh4QF9/b8/MyqVShQUFLS43YaGBnz00UdYu3atum3GjBlYuHAhli1bBj09PeTn52Ps2LHqwuufysvLcfbsWQQEBLTpeJozYsQI2fuqqiosXboUu3fvxrlz51BfX4+rV6/i1KnGmb/z8/Ohr6+PcePGNbs9BwcHTJ48GUlJSfD19cVXX32F2tpaPPTQQ7c91pbwMkk3IEkS9A1ad38zERERkaZJkgRDyVAnr7ZOODZr1ixs2bIFlZWVSE5OxqBBg9Qf1t944w2sXbsWixYtQlZWFvLz8xEUFIRr1651WKzy8vIQGRmJSZMmYdu2bThy5AiWLFnSofv4pxsLJEmSoFKpWuy/a9cunDlzBuHh4TAwMICBgQEiIiJQVlaGPXv2AABMTExaXP9mywBAT6+xHPnns3otPYP2z8ISAJ577jls27YNq1evxr59+5Cfnw9PT0917G61bwCYPXs20tLScPXqVSQnJyM8PFyjE6aw+CIiIiKiHmvatGnQ09PDpk2bsGHDBsycOVNdwOXm5mLKlCmYMWMGvL294ezsjF9//bXV23Z3d8eZM2dw7tw5dduBAwdkffbv3w9HR0csWbIEI0aMgKurK8rKymR9jIyM0NDQcNN9DRkyBEePHkV1dbW6LTc3F3p6enBzc2v1mG+UmJiIiIgI5Ofny14RERHqiTe8vLywb9++Zosmc3NzODk5qQu1G9na2gKALEb/nHzjZvbv34+HH34YDzzwADw9PWFvb4+TJ0+ql3t6ekKlUiEnJ6fFbUyaNAlmZmZISEjAzp07MXPmzFbtu71YfBERERFRj9WrVy+Eh4fjhRdewLlz5xATE6Ne5urqiszMTOzfvx+FhYV47LHH8Pvvv7d624GBgXBxcUFMTAyOHj2Kffv2YcmSJbI+rq6uOHXqFNLS0lBSUoJ169YhIyND1sfJyQmlpaXIz8/HhQsXmv2ercjISBgbGyM6OhrHjx9HVlYW5s+fj6ioKPUth231xx9/4KuvvkJ0dDTuvPNO2euRRx7B1q1bcfHiRcTGxqKiogIRERH48ccfUVxcjI8//hhFRUUAGr+n7M0338S6detQXFyMw4cP45133gHQeHXqrrvuQlxcHAoLC5GTkyN7Bu5mXFxc8NVXXyE/Px9Hjx7Fww8/LLuK5+TkhOjoaMycORNbt25FaWkpsrOzkZ6eru6jr6+PmJgYvPDCC3B1dW32ttCOxOKLiIiIiHq0WbNm4a+//kJQUJDs+ayXXnoJw4cPR1BQEPz9/WFvb4+wsLBWb1dPTw8ff/wxampq4Ovri9mzZ+Pf//63rM/999+Pp59+GrGxsRg2bBj279+PpUuXyvpMnToVwcHBGD9+PGxtbZud7t7U1BS7du3CxYsXMXLkSDz44IMICAjAu+++27Zg/MP1yTuae14rICAAJiYm2LhxI/r06YO9e/eiqqoK48aNg4+PDz788EP1LY7R0dF4++238d5778HDwwP33XcfiouL1dtKSkpCfX09fHx8sGDBAqxevbpV43vzzTdhZWWFu+++G6GhoQgKCsLw4cNlfRISEvDggw/iySefhLu7O+bMmSO7Ogg0/ve/du0aHn300baGqM0k0dovQyCZiooKWFpa4vLly21+ILKj1dXVYfv27Zg0aVKzDzrS7WF8NYvx1SzGV7MYX81ifDWvI2JcU1OD0tJSDBw4EMbGxh08wq5NpVKhoqJCPRsfdayOiu++ffsQEBCA06dP3/Qq4c1yvbW1AWc7JCIiIiKiHqe2thZ//PEHVqxYgYceeqjdt2e2BUtwIiIiIiLqcVJTU+Ho6IhLly7h9ddf18o+WXwREREREVGPExMTg4aGBhw6dAj9+vXTyj5ZfBEREREREWkBiy8iIiIium2cw426u47IcRZfRERERNRu12dJvHLlio5HQqRZ13P8dmZf5WyHRERERNRu+vr6sLKyQnl5OYDG75uSJEnHo+ocVCoVrl27hpqaGk41rwHaiq8QAleuXEF5eTmsrKygr6/f7m2x+CIiIiKi22Jvbw8A6gKMGgkhcPXqVZiYmLAg1QBtx9fKykqd6+3F4ouIiIiIboskSVAqlejbty/q6up0PZxOo66uDt9++y3uueceflG4BmgzvoaGhrd1xes6Fl9ERERE1CH09fU75ANqd6Gvr4/6+noYGxuz+NKArhhf3nxKRERERESkBSy+iIiIiIiItIDFFxERERERkRbwma92uv4laxUVFToeSePDhleuXEFFRUWXud+1K2F8NYvx1SzGV7MYX81ifDWPMdYsxlezOlN8r9cEt/oiZhZf7VRZWQkAGDBggI5HQkREREREnUFlZSUsLS1bXC6JW5Vn1CyVSoWzZ8/C3Nxc59/bUFFRgQEDBuD06dOwsLDQ6Vi6I8ZXsxhfzWJ8NYvx1SzGV/MYY81ifDWrM8VXCIHKyko4ODjc9AufeeWrnfT09NC/f39dD0PGwsJC54nXnTG+msX4ahbjq1mMr2YxvprHGGsW46tZnSW+N7vidR0n3CAiIiIiItICFl9ERERERERawOKrG1AoFFi+fDkUCoWuh9ItMb6axfhqFuOrWYyvZjG+mscYaxbjq1ldMb6ccIOIiIiIiEgLeOWLiIiIiIhIC1h8ERERERERaQGLLyIiIiIiIi1g8UVERERERKQFLL46mW+//RahoaFwcHCAJEnYunXrLdfJzs7G8OHDoVAo4OLigpSUlCZ91q9fDycnJxgbG2PUqFE4ePBgxw++C2hrfD///HNMmDABtra2sLCwgJ+fH3bt2iXrs2LFCkiSJHu5u7tr8Cg6r7bGNzs7u0nsJEnC+fPnZf2Yv39ra4xjYmKajbGHh4e6D3O40auvvoqRI0fC3Nwcffv2RVhYGIqKim653ubNm+Hu7g5jY2N4enpi+/btsuVCCCxbtgxKpRImJiYIDAxEcXGxpg6j02pPfD/88EOMHTsW1tbWsLa2RmBgYJPf/+ZyPDg4WJOH0im1J74pKSlNYmdsbCzrw/xt1J74+vv7N3v+nTx5sroP87dRQkICvLy81F+W7Ofnhx07dtx0na567mXx1clUV1fD29sb69evb1X/0tJSTJ48GePHj0d+fj4WLFiA2bNnywqETz/9FM888wyWL1+Ow4cPw9vbG0FBQSgvL9fUYXRabY3vt99+iwkTJmD79u04dOgQxo8fj9DQUBw5ckTWz8PDA+fOnVO/vvvuO00Mv9Nra3yvKyoqksWvb9++6mXMX7m2xnjt2rWy2J4+fRq9e/fGQw89JOvHHAZycnIwb948HDhwAJmZmairq8PEiRNRXV3d4jr79+/H9OnTMWvWLBw5cgRhYWEICwvD8ePH1X1ef/11rFu3Du+//z6+//57mJmZISgoCDU1Ndo4rE6jPfHNzs7G9OnTkZWVhby8PAwYMAATJ07EmTNnZP2Cg4Nl+Zuamqrpw+l02hNfALCwsJDFrqysTLac+duoPfH9/PPPZbE9fvw49PX1m5x/mb9A//79ERcXh0OHDuHHH3/EvffeiylTpuCnn35qtn+XPvcK6rQAiIyMjJv2ef7554WHh4esLTw8XAQFBanf+/r6innz5qnfNzQ0CAcHB/Hqq6926Hi7mtbEtzlDhw4VK1euVL9fvny58Pb27riBdROtiW9WVpYAIP76668W+zB/W9aeHM7IyBCSJImTJ0+q25jDzSsvLxcARE5OTot9pk2bJiZPnixrGzVqlHjssceEEEKoVCphb28v3njjDfXyS5cuCYVCIVJTUzUz8C6iNfG9UX19vTA3NxcfffSRui06OlpMmTJFAyPs2loT3+TkZGFpadnicuZvy9qTv2vWrBHm5uaiqqpK3cb8bZm1tbX473//2+yyrnzu5ZWvLi4vLw+BgYGytqCgIOTl5QEArl27hkOHDsn66OnpITAwUN2HWk+lUqGyshK9e/eWtRcXF8PBwQHOzs6IjIzEqVOndDTCrmnYsGFQKpWYMGECcnNz1e3M346XmJiIwMBAODo6ytqZw01dvnwZAJr8vv/Trc7BpaWlOH/+vKyPpaUlRo0a1eNzuDXxvdGVK1dQV1fXZJ3s7Gz07dsXbm5ueOKJJ/Dnn3926Fi7otbGt6qqCo6OjhgwYECTKw3M35a1J38TExMREREBMzMzWTvzV66hoQFpaWmorq6Gn59fs3268rmXxVcXd/78edjZ2cna7OzsUFFRgatXr+LChQtoaGhots+Nz9XQrcXHx6OqqgrTpk1Tt40aNQopKSnYuXMnEhISUFpairFjx6KyslKHI+0alEol3n//fWzZsgVbtmzBgAED4O/vj8OHDwMA87eDnT17Fjt27MDs2bNl7czhplQqFRYsWIAxY8bgzjvvbLFfS+fg6/l5/V/msFxr43ujRYsWwcHBQfaBKjg4GBs2bMCePXvw2muvIScnByEhIWhoaNDE0LuE1sbXzc0NSUlJ+OKLL7Bx40aoVCqMHj0av/32GwDmb0vak78HDx7E8ePHm5x/mb9/KygoQK9evaBQKPD4448jIyMDQ4cObbZvVz73Guh070RdyKZNm7By5Up88cUXsmeSQkJC1D97eXlh1KhRcHR0RHp6OmbNmqWLoXYZbm5ucHNzU78fPXo0SkpKsGbNGnz88cc6HFn39NFHH8HKygphYWGyduZwU/PmzcPx48d75LNv2tCe+MbFxSEtLQ3Z2dmySSEiIiLUP3t6esLLywuDBg1CdnY2AgICOnTcXUVr4+vn5ye7sjB69GgMGTIE//nPf7Bq1SpND7PLak/+JiYmwtPTE76+vrJ25u/f3NzckJ+fj8uXL+Ozzz5DdHQ0cnJyWizAuipe+eri7O3t8fvvv8vafv/9d1hYWMDExAQ2NjbQ19dvto+9vb02h9qlpaWlYfbs2UhPT29ymftGVlZWGDx4ME6cOKGl0XUvvr6+6tgxfzuOEAJJSUmIioqCkZHRTfv29ByOjY3Ftm3bkJWVhf79+9+0b0vn4Ov5ef1f5vDf2hLf6+Lj4xEXF4dvvvkGXl5eN+3r7OwMGxsb5m8b4nudoaEh/vWvf6ljx/xtqj3xra6uRlpaWqv+mNWT89fIyAguLi7w8fHBq6++Cm9vb6xdu7bZvl353Mviq4vz8/PDnj17ZG2ZmZnqv2QZGRnBx8dH1kelUmHPnj0t3kdLcqmpqXj00UeRmpoqmx62JVVVVSgpKYFSqdTC6Lqf/Px8deyYvx0nJycHJ06caNX//HtqDgshEBsbi4yMDOzduxcDBw685Tq3OgcPHDgQ9vb2sj4VFRX4/vvve1wOtye+QOOMZatWrcLOnTsxYsSIW/b/7bff8OeffzJ/Wxnff2poaEBBQYE6dszfv91OfDdv3oza2lrMmDHjln17av42R6VSoba2ttllXfrcq9PpPqiJyspKceTIEXHkyBEBQLz11lviyJEjoqysTAghxOLFi0VUVJS6///+9z9hamoqnnvuOVFYWCjWr18v9PX1xc6dO9V90tLShEKhECkpKeLnn38Wc+fOFVZWVuL8+fNaPz5da2t8P/nkE2FgYCDWr18vzp07p35dunRJ3efZZ58V2dnZorS0VOTm5orAwEBhY2MjysvLtX58utbW+K5Zs0Zs3bpVFBcXi4KCAvHUU08JPT09sXv3bnUf5q9cW2N83YwZM8SoUaOa3SZzuNETTzwhLC0tRXZ2tuz3/cqVK+o+UVFRYvHixer3ubm5wsDAQMTHx4vCwkKxfPlyYWhoKAoKCtR94uLihJWVlfjiiy/EsWPHxJQpU8TAgQPF1atXtXp8utae+MbFxQkjIyPx2WefydaprKwUQjT+PixcuFDk5eWJ0tJSsXv3bjF8+HDh6uoqampqtH6MutSe+K5cuVLs2rVLlJSUiEOHDomIiAhhbGwsfvrpJ3Uf5m+j9sT3urvvvluEh4c3aWf+/m3x4sUiJydHlJaWimPHjonFixcLSZLEN998I4ToXudeFl+dzPWpt298RUdHCyEapyQdN25ck3WGDRsmjIyMhLOzs0hOTm6y3XfeeUfccccdwsjISPj6+ooDBw5o/mA6obbGd9y4cTftL0Tj1P5KpVIYGRmJfv36ifDwcHHixAntHlgn0db4vvbaa2LQoEHC2NhY9O7dW/j7+4u9e/c22S7z92/tOUdcunRJmJiYiA8++KDZbTKHGzUXVwCyc+q4ceNkv/9CCJGeni4GDx4sjIyMhIeHh/j6669ly1UqlVi6dKmws7MTCoVCBAQEiKKiIi0cUefSnvg6Ojo2u87y5cuFEEJcuXJFTJw4Udja2gpDQ0Ph6Ogo5syZ0yP/ONOe+C5YsEB9brWzsxOTJk0Shw8flm2X+duoveeHX375RQBQFxH/xPz928yZM4Wjo6MwMjIStra2IiAgQBaz7nTulYQQooMuohEREREREVEL+MwXERERERGRFrD4IiIiIiIi0gIWX0RERERERFrA4ouIiIiIiEgLWHwRERERERFpAYsvIiIiIiIiLWDxRUREREREpAUsvoiIiIiIiLSAxRcREZEWSJKErVu36noYRESkQyy+iIio24uJiYEkSU1ewcHBuh4aERH1IAa6HgAREZE2BAcHIzk5WdamUCh0NBoiIuqJeOWLiIh6BIVCAXt7e9nL2toaQOMtgQkJCQgJCYGJiQmcnZ3x2WefydYvKCjAvffeCxMTE/Tp0wdz585FVVWVrE9SUhI8PDygUCigVCoRGxsrW37hwgU88MADMDU1haurK7788kv1sr/++guRkZGwtbWFiYkJXF1dmxSLRETUtbH4IiIiArB06VJMnToVR48eRWRkJCIiIlBYWAgAqK6uRlBQEKytrfHDDz9g8+bN2L17t6y4SkhIwLx58zB37lwUFBTgyy+/hIuLi2wfK1euxLRp03Ds2DFMmjQJkZGRuHjxonr/P//8M3bs2IHCwkIkJCTAxsZGewEgIiKNk4QQQteDICIi0qSYmBhs3LgRxsbGsvYXX3wRL774IiRJwuOPP46EhAT1srvuugvDhw/He++9hw8//BCLFi3C6dOnYWZmBgDYvn07QkNDcfbsWdjZ2aFfv3549NFHsXr16mbHIEkSXnrpJaxatQpAY0HXq1cv7NixA8HBwbj//vthY2ODpKQkDUWBiIh0jc98ERFRjzB+/HhZcQUAvXv3Vv/s5+cnW+bn54f8/HwAQGFhIby9vdWFFwCMGTMGKpUKRUVFkCQJZ8+eRUBAwE3H4OXlpf7ZzMwMFhYWKC8vBwA88cQTmDp1Kg4fPoyJEyciLCwMo0ePbtexEhFR58Tii4iIegQzM7MmtwF2FBMTk1b1MzQ0lL2XJAkqlQoAEBISgrKyMmzfvh2ZmZkICAjAvHnzEB8f3+HjJSIi3eAzX0RERAAOHDjQ5P2QIUMAAEOGDMHRo0dRXV2tXp6bmws9PT24ubnB3NwcTk5O2LNnz22NwdbWFtHR0di4cSPefvttfPDBB7e1PSIi6lx45YuIiHqE2tpanD9/XtZmYGCgntRi8+bNGDFiBO6++2588sknOHjwIBITEwEAkZGRWL58OaKjo7FixQr88ccfmD9/PqKiomBnZwcAWLFiBR5//HH07dsXISEhqKysRG5uLubPn9+q8S1btgw+Pj7w8PBAbW0ttm3bpi7+iIioe2DxRUREPcLOnTuhVCplbW5ubvjll18ANM5EmJaWhieffBJKpRKpqakYOnQoAMDU1BS7du3CU089hZEjR8LU1BRTp07FW2+9pd5WdHQ0ampqsGbNGixcuBA2NjZ48MEHWz0+IyMjvPDCCzh58iRMTEwwduxYpKWldcCRExFRZ8HZDomIqMeTJAkZGRkICwvT9VCIiKgb4zNfREREREREWsDii4iIiIiISAv4zBcREfV4vAOfiIi0gVe+iIiIiIiItIDFFxERERERkRaw+CIiIiIiItICFl9ERERERERawOKLiIiIiIhIC1h8ERERERERaQGLLyIiIiIiIi1g8UVERERERKQF/wf+VP0zFWMAfQAAAABJRU5ErkJggg==\n",
"text/plain": [
"<Figure size 1000x600 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Plotting the training and validation accuracy.\n",
"train_val_plot.accuracy_plot(history3a, [\"SlateBlue\", \"LightGreen\"])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<caption><span style=\"font-weight: bold;\">Figure 6</span> Training and Validation accuracy for model 3.</caption>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Likewise figure 6 displays overfitting for the same reasons as above. Rather than increasing, the validation accuracy decreases, however it increases again at the 3rd epoch. Despite this, the model still achieves statistical power."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"## 5.6 The fourth model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The third model still performed pretty well. I want to further simplify the model for the last time. This time I will only run 1 epoch. Table 5 displays the hyperparameters / parameters that I will be using for this model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<table style=\"width: 700px\">\n",
" <caption><span style=\"font-weight: bold;\">Table 5</span> Model 4 hyperparameters / parameters.</caption>\n",
" <tr style=\"background-color: #ECE5FC;\">\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Number of Layers</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Units</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Activation</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Epochs</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Batch Size</th>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">1</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[1]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[\"sigmoid\"]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">1</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">512</td>\n",
" </tr>\n",
"</table>"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 5.6.1 Building the model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"I am using the `compile_fit_model()` function to create, compile, and fit the model."
]
},
{
"cell_type": "code",
"execution_count": 54,
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"2813/2813 [==============================] - 34s 12ms/step - loss: 0.4057 - accuracy: 0.8258 - val_loss: 0.3665 - val_accuracy: 0.8411\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Creating, compiling, and fitting the model\n",
"history4a = compile_fit_model(units=[1], \n",
" activation=[\"sigmoid\"], \n",
" num_of_layers=1,\n",
" epochs=1, \n",
" batch_size=512)"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 5.6.2 Summary of results"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Since only 1 epoch was run, there is not point plotting a graph for these results. Instead I will summarise the results in a table."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<table style=\"width: 700px\">\n",
" <caption><span style=\"font-weight: bold;\">Table 6</span> Model 4 Summary.</caption>\n",
" <tr style=\"background-color: #ECE5FC;\">\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Loss</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Validation Loss</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Accuracy</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Validation Accuracy</th>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">0.3669</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">0.3661</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">0.8415</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">0.8418</td>\n",
" </tr>\n",
"</table>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Table 6 above shows that the validation data did better than the training data, this decreases the likelihood of overfitting. It also proves that the model learnt something, since the validation improved upon the training data results. This model also achieved statistical power (reaching ~84% accuracy). It is not possible to further simplify the model, therefore this model is the simplist model that achieves statistical power."
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "wXS3zyR5NvZC",
"tags": []
},
"source": [
"# 6 Scaling up - developing a model that overfits"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Being that I have achieved statistical power I will now develop a model that overfits. To achieve this, I will increase my model capacity (by increasing the number of layers and units). The reason I am aiming to overfit is because the \"*universal tension in machine learning is between optimization and generalization*\"[6] and therefore the optimal model is one \"*that stands right at the border between underfitting and overfitting*\"[6]. Yet in order to determine that border, the model must first cross it.\n",
"\n",
"Considering that my model predicts product reviews anything better than common sense is enough (especially since a lot bias is involved). If my problem was bank fraud or something similar, I would want a model that is at least 99.9% accurate- otherwise I would be alarming lots of customers for no reason. That being said, I am still aiming to scale up, make the model more powerful, and get the best accuracy that I possibly can. Though for my problem domain 80% to 90% is sufficient."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"## 6.1 The first model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Since I am aiming to create a model that overfits I will create a larger model. In the previous section my models were either 1 layer or 2 layers (and yet they achieved statistical power), therefore I will begin with 3 layers. I will also increase the number of epochs to 10. Table 7 displays the hyperparameters / parameters I will be using for the model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<table>\n",
" <caption><span style=\"font-weight: bold;\">Table 7</span> Model 1 hyperparameters / parameters.</caption>\n",
" <tr style=\"background-color: #ECE5FC;\">\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Number of Layers</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Units</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Activation</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Epochs</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Batch Size</th>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">3</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[32, 16, 1]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[\"relu\", \"relu\", \"sigmoid\"]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">10</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">512</td>\n",
" </tr>\n",
"</table>"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 6.1.1 Building the model"
]
},
{
"cell_type": "code",
"execution_count": 43,
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/10\n",
"2813/2813 [==============================] - 57s 19ms/step - loss: 0.3544 - accuracy: 0.8426 - val_loss: 0.3289 - val_accuracy: 0.8556\n",
"Epoch 2/10\n",
"2813/2813 [==============================] - 46s 16ms/step - loss: 0.3234 - accuracy: 0.8581 - val_loss: 0.3192 - val_accuracy: 0.8608\n",
"Epoch 3/10\n",
"2813/2813 [==============================] - 44s 16ms/step - loss: 0.3139 - accuracy: 0.8634 - val_loss: 0.3165 - val_accuracy: 0.8621\n",
"Epoch 4/10\n",
"2813/2813 [==============================] - 46s 16ms/step - loss: 0.3082 - accuracy: 0.8665 - val_loss: 0.3127 - val_accuracy: 0.8640\n",
"Epoch 5/10\n",
"2813/2813 [==============================] - 46s 16ms/step - loss: 0.3041 - accuracy: 0.8687 - val_loss: 0.3107 - val_accuracy: 0.8655\n",
"Epoch 6/10\n",
"2813/2813 [==============================] - 36s 13ms/step - loss: 0.3011 - accuracy: 0.8704 - val_loss: 0.3112 - val_accuracy: 0.8648\n",
"Epoch 7/10\n",
"2813/2813 [==============================] - 30s 11ms/step - loss: 0.2990 - accuracy: 0.8713 - val_loss: 0.3113 - val_accuracy: 0.8651\n",
"Epoch 8/10\n",
"2813/2813 [==============================] - 42s 15ms/step - loss: 0.2971 - accuracy: 0.8725 - val_loss: 0.3100 - val_accuracy: 0.8655\n",
"Epoch 9/10\n",
"2813/2813 [==============================] - 39s 14ms/step - loss: 0.2956 - accuracy: 0.8731 - val_loss: 0.3092 - val_accuracy: 0.8658\n",
"Epoch 10/10\n",
"2813/2813 [==============================] - 39s 14ms/step - loss: 0.2944 - accuracy: 0.8738 - val_loss: 0.3092 - val_accuracy: 0.8659\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Creating, compiling, and fitting the model\n",
"history1b = compile_fit_model(units=[32, 16, 1], \n",
" activation=[\"relu\", \"relu\", \"sigmoid\"], \n",
" num_of_layers=3,\n",
" epochs=10, \n",
" batch_size=512)"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 6.1.2 Plotting the training and validation loss"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Just like in the previous section, I will plot the training and validation loss so that I can make comparisons. I will use the instance of the `TrainValPlot` class from before. Then I will call the `loss_plot()` method."
]
},
{
"cell_type": "code",
"execution_count": 46,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 1000x600 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Plotting the training and validation loss.\n",
"train_val_plot.loss_plot(history1b)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<caption><span style=\"font-weight: bold;\">Figure 7</span> Training and Validation loss for model 1.</caption>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Based of figure 7 it seems that overfitting occurs around the 4th epoch, since this is where the training loss and validation loss begin to diverge. In addition, from then on the training loss decreases rapidly whereas the validation loss begins to plateau."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 6.1.3 Plotting the training and validation accuracy"
]
},
{
"cell_type": "code",
"execution_count": 47,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 1000x600 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Plotting the training and validation accuracy.\n",
"train_val_plot.accuracy_plot(history1b, [\"SlateBlue\", \"LightGreen\"])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<caption><span style=\"font-weight: bold;\">Figure 8</span> Training and Validation accuracy for model 1.</caption>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Figure 8 behaves like figure 7. At the 4th epoch overfitting begins to take place, as made obvious by the divergence of the green (validation) and purple (training) line. The training accuracy reaches approximately 87.3% whereas the validation accuracy reaches approximately 86.5%. Although the graph displays overfitting, the differences between the two accuracies are not that high. If I were to just look at the numbers, I might not have realised that overfitting had taken place."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"## 6.2 The second model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Although the previous model overfitted and reached an accuracy of about 87%, I would like to build a bigger model and see whether the loss can decrease and the accuracy can increase. In addition, I will increase the number of epochs to 15. Table 8 displays the hyperparameters / parameters I will be using for the model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<table>\n",
" <caption><span style=\"font-weight: bold;\">Table 8</span> Model 2 hyperparameters / parameters.</caption>\n",
" <tr style=\"background-color: #ECE5FC;\">\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Number of Layers</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Units</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Activation</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Epochs</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Batch Size</th>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">4</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[64, 32, 16, 1]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[\"relu\", \"relu\", \"relu\", \"sigmoid\"]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">15</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">512</td>\n",
" </tr>\n",
"</table>"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 6.2.1 Building the model"
]
},
{
"cell_type": "code",
"execution_count": 48,
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/15\n",
"2813/2813 [==============================] - 45s 15ms/step - loss: 0.3449 - accuracy: 0.8473 - val_loss: 0.3189 - val_accuracy: 0.8610\n",
"Epoch 2/15\n",
"2813/2813 [==============================] - 48s 17ms/step - loss: 0.3109 - accuracy: 0.8651 - val_loss: 0.3095 - val_accuracy: 0.8662\n",
"Epoch 3/15\n",
"2813/2813 [==============================] - 41s 15ms/step - loss: 0.2995 - accuracy: 0.8711 - val_loss: 0.3069 - val_accuracy: 0.8671\n",
"Epoch 4/15\n",
"2813/2813 [==============================] - 37s 13ms/step - loss: 0.2921 - accuracy: 0.8753 - val_loss: 0.3038 - val_accuracy: 0.8691\n",
"Epoch 5/15\n",
"2813/2813 [==============================] - 36s 13ms/step - loss: 0.2871 - accuracy: 0.8777 - val_loss: 0.3035 - val_accuracy: 0.8692\n",
"Epoch 6/15\n",
"2813/2813 [==============================] - 42s 15ms/step - loss: 0.2831 - accuracy: 0.8798 - val_loss: 0.3040 - val_accuracy: 0.8690\n",
"Epoch 7/15\n",
"2813/2813 [==============================] - 44s 16ms/step - loss: 0.2799 - accuracy: 0.8813 - val_loss: 0.3039 - val_accuracy: 0.8693\n",
"Epoch 8/15\n",
"2813/2813 [==============================] - 47s 17ms/step - loss: 0.2775 - accuracy: 0.8827 - val_loss: 0.3039 - val_accuracy: 0.8694\n",
"Epoch 9/15\n",
"2813/2813 [==============================] - 46s 16ms/step - loss: 0.2751 - accuracy: 0.8838 - val_loss: 0.3049 - val_accuracy: 0.8691\n",
"Epoch 10/15\n",
"2813/2813 [==============================] - 47s 17ms/step - loss: 0.2733 - accuracy: 0.8848 - val_loss: 0.3068 - val_accuracy: 0.8681\n",
"Epoch 11/15\n",
"2813/2813 [==============================] - 42s 15ms/step - loss: 0.2717 - accuracy: 0.8856 - val_loss: 0.3068 - val_accuracy: 0.8684\n",
"Epoch 12/15\n",
"2813/2813 [==============================] - 30s 11ms/step - loss: 0.2701 - accuracy: 0.8866 - val_loss: 0.3070 - val_accuracy: 0.8683\n",
"Epoch 13/15\n",
"2813/2813 [==============================] - 29s 10ms/step - loss: 0.2688 - accuracy: 0.8872 - val_loss: 0.3092 - val_accuracy: 0.8678\n",
"Epoch 14/15\n",
"2813/2813 [==============================] - 43s 15ms/step - loss: 0.2677 - accuracy: 0.8879 - val_loss: 0.3086 - val_accuracy: 0.8679\n",
"Epoch 15/15\n",
"2813/2813 [==============================] - 46s 16ms/step - loss: 0.2665 - accuracy: 0.8885 - val_loss: 0.3094 - val_accuracy: 0.8673\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Creating, compiling, and fitting the model\n",
"history2b = compile_fit_model(units=[64, 32, 16, 1], \n",
" activation=[\"relu\", \"relu\", \"relu\", \"sigmoid\"], \n",
" num_of_layers=4,\n",
" epochs=15, \n",
" batch_size=512)"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 6.2.2 Plotting the training and validation loss"
]
},
{
"cell_type": "code",
"execution_count": 49,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 1000x600 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Plotting the training and validation loss.\n",
"train_val_plot.loss_plot(history2b)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<caption><span style=\"font-weight: bold;\">Figure 9</span> Training and Validation loss for model 2.</caption>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Figure 9 shows that overfitting definitely occurs. It seems to begin again around the 4th epoch. From then on, the validation loss plateaus and then increases, whereas the training loss continues to drop. This makes it clear that this model doesn't generalise well to unseen data."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 6.2.3 Plotting the training and validation accuracy"
]
},
{
"cell_type": "code",
"execution_count": 50,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 1000x600 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Plotting the training and validation accuracy.\n",
"train_val_plot.accuracy_plot(history2b, [\"SlateBlue\", \"LightGreen\"])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<caption><span style=\"font-weight: bold;\">Figure 10</span> Training and Validation accuracy for model 2.</caption>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The same is true for figure 10 that also displays overfitting (starting at the 4th epoch). Looking at the trajectory of the green line (validation accuracy) tells us that the accuracy begins to decrease after it plateaus. On the other hand, the training accuracy reaches approximately 88.8% and the validation accuracy reaches approximately 86.7%, meaning that although overfitting occurs, they are not too far off from each other. Furthermore, 88.8% is the highest accuracy that a model has produced so far."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"## 6.3 The third model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The last model overfitted and reached an accuracy of approximately 88.8%- which is quite good. However, I don't believe that my model is that complex. For this next model, I will increase the number of units, while keeping the number of layers the same. I will also increase the epochs to 20, and decrease the batch size to 128. Table 9 displays the hyperparameters / parameters I will be using for the model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<table>\n",
" <caption><span style=\"font-weight: bold;\">Table 9</span> Model 3 hyperparameters / parameters.</caption>\n",
" <tr style=\"background-color: #ECE5FC;\">\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Number of Layers</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Units</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Activation</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Epochs</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Batch Size</th>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">4</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[128, 64, 32, 1]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[\"relu\", \"relu\", \"relu\", \"sigmoid\"]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">20</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">128</td>\n",
" </tr>\n",
"</table>"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 6.3.1 Building the model"
]
},
{
"cell_type": "code",
"execution_count": 51,
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/20\n",
"11250/11250 [==============================] - 123s 11ms/step - loss: 0.3340 - accuracy: 0.8542 - val_loss: 0.3165 - val_accuracy: 0.8641\n",
"Epoch 2/20\n",
"11250/11250 [==============================] - 94s 8ms/step - loss: 0.3096 - accuracy: 0.8685 - val_loss: 0.3102 - val_accuracy: 0.8677\n",
"Epoch 3/20\n",
"11250/11250 [==============================] - 108s 10ms/step - loss: 0.2999 - accuracy: 0.8743 - val_loss: 0.3074 - val_accuracy: 0.8694\n",
"Epoch 4/20\n",
"11250/11250 [==============================] - 111s 10ms/step - loss: 0.2926 - accuracy: 0.8783 - val_loss: 0.3081 - val_accuracy: 0.8700\n",
"Epoch 5/20\n",
"11250/11250 [==============================] - 104s 9ms/step - loss: 0.2869 - accuracy: 0.8818 - val_loss: 0.3079 - val_accuracy: 0.8692\n",
"Epoch 6/20\n",
"11250/11250 [==============================] - 92s 8ms/step - loss: 0.2827 - accuracy: 0.8845 - val_loss: 0.3076 - val_accuracy: 0.8691\n",
"Epoch 7/20\n",
"11250/11250 [==============================] - 95s 8ms/step - loss: 0.2786 - accuracy: 0.8869 - val_loss: 0.3097 - val_accuracy: 0.8692\n",
"Epoch 8/20\n",
"11250/11250 [==============================] - 126s 11ms/step - loss: 0.2757 - accuracy: 0.8889 - val_loss: 0.3122 - val_accuracy: 0.8677\n",
"Epoch 9/20\n",
"11250/11250 [==============================] - 125s 11ms/step - loss: 0.2729 - accuracy: 0.8904 - val_loss: 0.3112 - val_accuracy: 0.8679\n",
"Epoch 10/20\n",
"11250/11250 [==============================] - 128s 11ms/step - loss: 0.2706 - accuracy: 0.8921 - val_loss: 0.3182 - val_accuracy: 0.8661\n",
"Epoch 11/20\n",
"11250/11250 [==============================] - 116s 10ms/step - loss: 0.2684 - accuracy: 0.8933 - val_loss: 0.3152 - val_accuracy: 0.8662\n",
"Epoch 12/20\n",
"11250/11250 [==============================] - 91s 8ms/step - loss: 0.2667 - accuracy: 0.8945 - val_loss: 0.3160 - val_accuracy: 0.8658\n",
"Epoch 13/20\n",
"11250/11250 [==============================] - 493s 44ms/step - loss: 0.2649 - accuracy: 0.8958 - val_loss: 0.3202 - val_accuracy: 0.8659\n",
"Epoch 14/20\n",
"11250/11250 [==============================] - 74s 7ms/step - loss: 0.2634 - accuracy: 0.8968 - val_loss: 0.3195 - val_accuracy: 0.8654\n",
"Epoch 15/20\n",
"11250/11250 [==============================] - 165s 15ms/step - loss: 0.2617 - accuracy: 0.8979 - val_loss: 0.3275 - val_accuracy: 0.8647\n",
"Epoch 16/20\n",
"11250/11250 [==============================] - 81s 7ms/step - loss: 0.2600 - accuracy: 0.8987 - val_loss: 0.3261 - val_accuracy: 0.8650\n",
"Epoch 17/20\n",
"11250/11250 [==============================] - 88s 8ms/step - loss: 0.2585 - accuracy: 0.8997 - val_loss: 0.3227 - val_accuracy: 0.8638\n",
"Epoch 18/20\n",
"11250/11250 [==============================] - 82s 7ms/step - loss: 0.2571 - accuracy: 0.9006 - val_loss: 0.3288 - val_accuracy: 0.8638\n",
"Epoch 19/20\n",
"11250/11250 [==============================] - 87s 8ms/step - loss: 0.2561 - accuracy: 0.9012 - val_loss: 0.3291 - val_accuracy: 0.8644\n",
"Epoch 20/20\n",
"11250/11250 [==============================] - 80s 7ms/step - loss: 0.2551 - accuracy: 0.9019 - val_loss: 0.3297 - val_accuracy: 0.8625\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Creating, compiling, and fitting the model\n",
"history3b = compile_fit_model(units=[128, 64, 32, 1], \n",
" activation=[\"relu\", \"relu\", \"relu\", \"sigmoid\"], \n",
" num_of_layers=4,\n",
" epochs=20, \n",
" batch_size=128)"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 6.3.2 Plotting the training and validation loss"
]
},
{
"cell_type": "code",
"execution_count": 52,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 1000x600 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Plotting the training and validation loss.\n",
"train_val_plot.loss_plot(history3b)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<caption><span style=\"font-weight: bold;\">Figure 11</span> Training and Validation loss for model 3.</caption>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Figure 11 displays that this model also overfits, occurring from around the 3rd epoch. It seems like the more complex the model, the earlier overfitting occurs- though it is too early to tell, and the problem type and dataset size will also bear some influence. From the 3rd epoch, the validation loss becomes noisy and erratic, which is another clear indicator of overfitting data."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 6.3.3 Plotting the training and validation accuracy"
]
},
{
"cell_type": "code",
"execution_count": 53,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 1000x600 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Plotting the training and validation accuracy.\n",
"train_val_plot.accuracy_plot(history3b, [\"SlateBlue\", \"LightGreen\"])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<caption><span style=\"font-weight: bold;\">Figure 12</span> Training and Validation accuracy for model 3.</caption>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Just like in figure 11, figure 12 also displays overfitting occurring from about the 3rd epoch. The validation accuracy reaches a high of approximately 87%, but ends around 86.3%. On the other hand, the training accuracy ends around 90.2%, which is almost 4% greater. I believe that this is the largest gap that a model has displayed so far. This makes a lot of sense, since it is the most complex model so far."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"## 6.4 The fourth model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"I have only been steadily increasing my model so far, which is unnecessary since I am not at the point of experimentation or tuning. The goal is to build the most powerful model I can, and I am trying to push my model to the brink. Therefore, I will increase the number of layers to 7, while keeping the number of epochs at 20. Table 10 displays the hyperparameters / parameters I will be using for the model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<table>\n",
" <caption><span style=\"font-weight: bold;\">Table 10</span> Model 4 hyperparameters / parameters.</caption>\n",
" <tr style=\"background-color: #ECE5FC;\">\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Number of Layers</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Units</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Activation</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Epochs</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Batch Size</th>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">7</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[256, 128, 128, 128, 64, 32, 1]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[\"relu\", \"relu\", \"relu\", \"relu\", \"relu\",\n",
" \"relu\", \"sigmoid\"]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">20</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">512</td>\n",
" </tr>\n",
"</table>"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 6.4.1 Building the model"
]
},
{
"cell_type": "code",
"execution_count": 54,
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/20\n",
"2813/2813 [==============================] - 118s 40ms/step - loss: 0.3375 - accuracy: 0.8505 - val_loss: 0.3220 - val_accuracy: 0.8592\n",
"Epoch 2/20\n",
"2813/2813 [==============================] - 92s 33ms/step - loss: 0.2954 - accuracy: 0.8734 - val_loss: 0.3021 - val_accuracy: 0.8699\n",
"Epoch 3/20\n",
"2813/2813 [==============================] - 80s 29ms/step - loss: 0.2767 - accuracy: 0.8831 - val_loss: 0.2980 - val_accuracy: 0.8723\n",
"Epoch 4/20\n",
"2813/2813 [==============================] - 92s 33ms/step - loss: 0.2628 - accuracy: 0.8902 - val_loss: 0.3070 - val_accuracy: 0.8714\n",
"Epoch 5/20\n",
"2813/2813 [==============================] - 77s 27ms/step - loss: 0.2510 - accuracy: 0.8962 - val_loss: 0.3028 - val_accuracy: 0.8698\n",
"Epoch 6/20\n",
"2813/2813 [==============================] - 79s 28ms/step - loss: 0.2402 - accuracy: 0.9021 - val_loss: 0.3092 - val_accuracy: 0.8691\n",
"Epoch 7/20\n",
"2813/2813 [==============================] - 74s 26ms/step - loss: 0.2308 - accuracy: 0.9067 - val_loss: 0.3180 - val_accuracy: 0.8675\n",
"Epoch 8/20\n",
"2813/2813 [==============================] - 73s 26ms/step - loss: 0.2219 - accuracy: 0.9112 - val_loss: 0.3174 - val_accuracy: 0.8661\n",
"Epoch 9/20\n",
"2813/2813 [==============================] - 74s 26ms/step - loss: 0.2141 - accuracy: 0.9154 - val_loss: 0.3375 - val_accuracy: 0.8631\n",
"Epoch 10/20\n",
"2813/2813 [==============================] - 73s 26ms/step - loss: 0.2067 - accuracy: 0.9191 - val_loss: 0.3313 - val_accuracy: 0.8629\n",
"Epoch 11/20\n",
"2813/2813 [==============================] - 72s 26ms/step - loss: 0.1998 - accuracy: 0.9224 - val_loss: 0.3453 - val_accuracy: 0.8618\n",
"Epoch 12/20\n",
"2813/2813 [==============================] - 73s 26ms/step - loss: 0.1938 - accuracy: 0.9255 - val_loss: 0.3579 - val_accuracy: 0.8605\n",
"Epoch 13/20\n",
"2813/2813 [==============================] - 78s 28ms/step - loss: 0.1882 - accuracy: 0.9283 - val_loss: 0.3613 - val_accuracy: 0.8591\n",
"Epoch 14/20\n",
"2813/2813 [==============================] - 74s 26ms/step - loss: 0.1827 - accuracy: 0.9310 - val_loss: 0.3615 - val_accuracy: 0.8581\n",
"Epoch 15/20\n",
"2813/2813 [==============================] - 84s 30ms/step - loss: 0.1777 - accuracy: 0.9335 - val_loss: 0.3794 - val_accuracy: 0.8565\n",
"Epoch 16/20\n",
"2813/2813 [==============================] - 87s 31ms/step - loss: 0.1734 - accuracy: 0.9355 - val_loss: 0.3925 - val_accuracy: 0.8563\n",
"Epoch 17/20\n",
"2813/2813 [==============================] - 114s 41ms/step - loss: 0.1686 - accuracy: 0.9377 - val_loss: 0.4021 - val_accuracy: 0.8558\n",
"Epoch 18/20\n",
"2813/2813 [==============================] - 115s 41ms/step - loss: 0.1645 - accuracy: 0.9399 - val_loss: 0.4007 - val_accuracy: 0.8542\n",
"Epoch 19/20\n",
"2813/2813 [==============================] - 104s 37ms/step - loss: 0.1607 - accuracy: 0.9416 - val_loss: 0.4004 - val_accuracy: 0.8534\n",
"Epoch 20/20\n",
"2813/2813 [==============================] - 104s 37ms/step - loss: 0.1569 - accuracy: 0.9435 - val_loss: 0.4215 - val_accuracy: 0.8526\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Creating, compiling, and fitting the model\n",
"history4b = compile_fit_model(units=[256, 128, 128, 128, 64, 32, 1], \n",
" activation=[\"relu\", \"relu\", \"relu\", \"relu\", \n",
" \"relu\", \"relu\", \"sigmoid\"],\n",
" num_of_layers=7,\n",
" epochs=20, \n",
" batch_size=512)"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 6.4.2 Plotting the training and validation loss"
]
},
{
"cell_type": "code",
"execution_count": 55,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 1000x600 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Plotting the training and validation loss.\n",
"train_val_plot.loss_plot(history4b)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<caption><span style=\"font-weight: bold;\">Figure 13</span> Training and Validation loss for model 4.</caption>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Figure 13 shows that overfitting occurs almost instantly (from the 2nd epoch). The validation loss only decreases from the 1st epoch to the 2nd, but then from there on it becomes noisy and erratic, though it generally increases. Figure 13 displays a large divergence between the training and validation loss, with the validation loss ending higher than it started. This is unsurprising since larger more complex models are more susceptible to overfitting, especially when the number of epochs is high."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 6.4.3 Plotting the training and validation accuracy"
]
},
{
"cell_type": "code",
"execution_count": 56,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 1000x600 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Plotting the training and validation accuracy.\n",
"train_val_plot.accuracy_plot(history4b, [\"SlateBlue\", \"LightGreen\"])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<caption><span style=\"font-weight: bold;\">Figure 14</span> Training and Validation accuracy for model 4.</caption>"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Figure 14 shows that the training accuracy reaches 94% whereas the validation accuracy reaches approximately 85.3%. This gap is even larger than the last model. Overfitting occurs from the 2nd epoch, however this is the highest training accuracy reached so far."
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"## 6.5 The fifth model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The final model within this section will be a 10 layer model- 3 layers larger than the last, though I will decrease the number of epochs to 15. Table 11 displays the hyperparameters / parameters I will be using for the model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<table style=\"width: 850px;\">\n",
" <caption><span style=\"font-weight: bold;\">Table 11</span> Model 5 hyperparameters / parameters.</caption>\n",
" <tr style=\"background-color: #ECE5FC;\">\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Number of Layers</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Units</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Activation</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Epochs</th>\n",
" <th style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">Batch Size</th>\n",
" </tr>\n",
" <tr style=\"background-color: #FFFFFF;\">\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">10</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[512, 512, 256, 256, 128, 128, 128, 64, 32, 1]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">[\"relu\", \"relu\", \"relu\", \"relu\", \"relu\",\n",
" \"relu\", \"relu\", \"relu\", \"relu\", \"sigmoid\"]</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">15</td>\n",
" <td style=\"border: 1px solid #dddddd; text-align: center; padding: 8px;\">512</td>\n",
" </tr>\n",
"</table>"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 6.5.1 Building the model"
]
},
{
"cell_type": "code",
"execution_count": 43,
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/15\n",
"2813/2813 [==============================] - 307s 108ms/step - loss: 0.3401 - accuracy: 0.8493 - val_loss: 0.3068 - val_accuracy: 0.8668\n",
"Epoch 2/15\n",
"2813/2813 [==============================] - 268s 95ms/step - loss: 0.2916 - accuracy: 0.8753 - val_loss: 0.2990 - val_accuracy: 0.8724\n",
"Epoch 3/15\n",
"2813/2813 [==============================] - 351s 125ms/step - loss: 0.2687 - accuracy: 0.8869 - val_loss: 0.2979 - val_accuracy: 0.8726\n",
"Epoch 4/15\n",
"2813/2813 [==============================] - 289s 103ms/step - loss: 0.2489 - accuracy: 0.8971 - val_loss: 0.3038 - val_accuracy: 0.8687\n",
"Epoch 5/15\n",
"2813/2813 [==============================] - 260s 92ms/step - loss: 0.2305 - accuracy: 0.9063 - val_loss: 0.3078 - val_accuracy: 0.8688\n",
"Epoch 6/15\n",
"2813/2813 [==============================] - 291s 104ms/step - loss: 0.2130 - accuracy: 0.9148 - val_loss: 0.3272 - val_accuracy: 0.8666\n",
"Epoch 7/15\n",
"2813/2813 [==============================] - 234s 83ms/step - loss: 0.1961 - accuracy: 0.9230 - val_loss: 0.3417 - val_accuracy: 0.8645\n",
"Epoch 8/15\n",
"2813/2813 [==============================] - 271s 97ms/step - loss: 0.1802 - accuracy: 0.9306 - val_loss: 0.3712 - val_accuracy: 0.8617\n",
"Epoch 9/15\n",
"2813/2813 [==============================] - 316s 112ms/step - loss: 0.1649 - accuracy: 0.9376 - val_loss: 0.4035 - val_accuracy: 0.8571\n",
"Epoch 10/15\n",
"2813/2813 [==============================] - 340s 121ms/step - loss: 0.1509 - accuracy: 0.9440 - val_loss: 0.3958 - val_accuracy: 0.8571\n",
"Epoch 11/15\n",
"2813/2813 [==============================] - 289s 103ms/step - loss: 0.1371 - accuracy: 0.9501 - val_loss: 0.4199 - val_accuracy: 0.8559\n",
"Epoch 12/15\n",
"2813/2813 [==============================] - 299s 106ms/step - loss: 0.1251 - accuracy: 0.9551 - val_loss: 0.4329 - val_accuracy: 0.8531\n",
"Epoch 13/15\n",
"2813/2813 [==============================] - 232s 83ms/step - loss: 0.1138 - accuracy: 0.9598 - val_loss: 0.4731 - val_accuracy: 0.8513\n",
"Epoch 14/15\n",
"2813/2813 [==============================] - 248s 88ms/step - loss: 0.1034 - accuracy: 0.9639 - val_loss: 0.5275 - val_accuracy: 0.8510\n",
"Epoch 15/15\n",
"2813/2813 [==============================] - 229s 81ms/step - loss: 0.0938 - accuracy: 0.9676 - val_loss: 0.5489 - val_accuracy: 0.8490\n"
]
}
],
"source": [
"# I wrote all the code in this cell.\n",
"# Creating, compiling, and fitting the model\n",
"history5b = compile_fit_model(units=[512, 512, 256, 256, 128, 128, 128, 64,\n",
" 32, 1], \n",
" activation=[\"relu\", \"relu\", \"relu\", \"relu\", \n",
" \"relu\", \"relu\", \"relu\", \"relu\", \n",
" \"relu\", \"sigmoid\"],\n",
" num_of_layers=10,\n",
" epochs=15, \n",
" batch_size=512)"
]
},
{
"cell_type": "markdown",
"metadata": {
"tags": []
},
"source": [
"### 6.5.2 Plotting the training and validation loss"
]
},
{
"cell_type": "code",
"execution_count": 44,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment