sample.py

import numpy as np

predictions = [
    ('company_a:user1', 'model_a', 0.5),
    ('company_a:user1', 'model_b', 0.6),
    ('company_a:user1', 'model_c', 0.3),
    ('company_a:user2', 'model_a', 0.33),
    ('company_a:user2', 'model_b', 0.27),
    ('company_a:user2', 'model_c', 0.75),
    ('company_b:user1', 'model_a', 0.36),
    ('company_b:user1', 'model_b', 0.67),
    ('company_b:user1', 'model_c', 0.84),
    ('company_b:user2', 'model_a', 0.35),
    ('company_b:user2', 'model_b', 0.56),
    ('company_b:user2', 'model_c', 0.54),
    ('company_b:user3', 'model_a', 0.32),
    ('company_b:user3', 'model_b', 0.51)
]

pageviews = [
    ('company_a', 'user1', 'PAGE_TYPE_A'),
    ('company_a', 'user1', 'PAGE_TYPE_B'),
    ('company_a', 'user1', 'PAGE_TYPE_C'),
    ('company_a', 'user1', 'PAGE_TYPE_B'),
    ('company_a', 'user1', 'PAGE_TYPE_A'),
    ('company_a', 'user1', 'PAGE_TYPE_B'),
    ('company_a', 'user1', 'PAGE_TYPE_B'),
    ('company_a', 'user1', 'BUYSUCCESS'),
    ('company_a', 'user2', 'PAGE_TYPE_D'),
    ('company_a', 'user2', 'PAGE_TYPE_F'),
    ('company_a', 'user2', 'PAGE_TYPE_H'),
    ('company_a', 'user2', 'PAGE_TYPE_A'),
    ('company_a', 'user2', 'PAGE_TYPE_B'),
    ('company_a', 'user2', 'BUYSUCCESS'),
    ('company_a', 'user2', 'BUYSUCCESS'),
    ('company_a', 'user2', 'BUYSUCCESS'),
    ('company_b', 'user1', 'PAGE_TYPE_A'),
    ('company_b', 'user1', 'PAGE_TYPE_B'),
    ('company_b', 'user1', 'PAGE_TYPE_C'),
    ('company_b', 'user1', 'PAGE_TYPE_B'),
    ('company_b', 'user1', 'PAGE_TYPE_A'),
    ('company_b', 'user1', 'PAGE_TYPE_B'),
    ('company_b', 'user1', 'PAGE_TYPE_B'),
    ('company_b', 'user1', 'BUYSUCCESS'),
    ('company_b', 'user2', 'PAGE_TYPE_D'),
    ('company_b', 'user2', 'PAGE_TYPE_F'),
    ('company_b', 'user2', 'PAGE_TYPE_H'),
    ('company_b', 'user2', 'PAGE_TYPE_A'),
    ('company_b', 'user2', 'PAGE_TYPE_B'),
    ('company_b', 'user3', 'PAGE_TYPE_D'),
    ('company_b', 'user3', 'PAGE_TYPE_F'),
    ('company_b', 'user3', 'PAGE_TYPE_H'),
    ('company_b', 'user3', 'BUYSUCCESS'),
    ('company_b', 'user3', 'PAGE_TYPE_B')
]


companies = ['company_a', 'company_b', 'company_c']
models = ['model_a', 'model_b', 'model_c']
users = ['user1', 'user2', 'user3']
print("company, model, segment, segment_users, really_bought")
for company in companies:
    for model in models:
        for segment in np.linspace(0, 1, 10, endpoint=False):
            segment_users = 0
            really_bought = 0
            min_v = segment
            max_v = segment + 0.1

            users = []

            for pred in predictions:
                if pred[2] >= min_v and pred[2] < max_v and model == pred[1] and company == pred[0].split(':')[0]:
                    segment_users += 1
                    users.append(pred[0].split(':')[1])
            really_bought = 0
            for user in users:
                for pv in pageviews:
                    if company == pv[0] and user == pv[1] and pv[2] == 'BUYSUCCESS':
                        really_bought += 1

            if segment_users > 0:
                print(f"{company}, {model}, {min_v:.1f}-{max_v:.1f}, "
                       f"{segment_users}, {really_bought}")