bootstrap-evalute-test-conversions-difference.py

import pandas as pd
import numpy as np


raw_test_data = pd.read_csv('medium_ppc_inc_evaluate_dataset.csv')
# You can access the csv above in this link: https://drive.google.com/open?id=1IOuneJr-QFDYGsJPRjQ8ra-5C4VczeOW

raw_test_data['diff_conversions'] = raw_test_data['test_conversions'] - raw_test_data['ctl_conversions']

N = len(raw_test_data)
sd_mult = 1.64485
boot_array = []
for x in range(10000):
    boot_sample = raw_test_data.sample(n=N, replace=True)
    boot_sample_mean = float(boot_sample['diff_conversions'].sum()) / float(boot_sample['ctl_conversions'].sum())
    boot_array.append(boot_sample_mean)

boot_mean = np.mean(boot_array)

boot_mean_std = np.std(boot_array)

CI_90_B = (round(boot_mean - (sd_mult * boot_mean_std), 2),
           round(boot_mean + (sd_mult * boot_mean_std), 2))

print("point estimator: " + str(boot_mean))
print("std_dev: " + str(boot_mean_std))
print("CI: " + str(CI_90_B))