Generate pseudo logs for BS test

logs_generator.py

from scipy import stats
import pandas as pd

randgen = lambda n: [1 if i > .5 else 0 for i in stats.gamma.rvs(1, scale=0.7, size=n) > .5]


def truncate(l, n):
    try:
        ls = "".join(map(str, l))
        idx = ls.index("".join(map(str, [0] * n)))
        return l[:idx] + [0] * (len(l) - idx)
    except ValueError:
        return l


def generate_logs(days=90, users=1000, max_lazyness=5):
    d = {i: truncate(randgen(days), max_lazyness) for i in range(users)}
    logs = pd.DataFrame.from_dict(d).transpose()
    logs.columns = ['Day ' + str(i) for i in range(1, days + 1)]
    logs.index.name = "User ID"
    return logs


generate_logs().to_csv("logs.csv")