reduce size of CSV with pandas

sample.py

import pandas as pd


# maintain the original data format by reading everything as strings
original = pd.read_csv("original.csv", dtype="object")

# set the random_state so it's reproducible. alternatively, can pass a `frac` to use a percentage.
sampled = original.sample(n=5000, random_state=1).sort_index()

# exclude the index so the columns match the orignal
sampled.to_csv("sampled.csv", index=False)