In [3]:

    

import pandas as pd
import scipy as sp
import daft

import matplotlib.pyplot as plt
import numpy as np

from abtools import ABtest, BernoulliModel

%matplotlib inline


    

In [4]:

    

p = .05
size = 200000
delta_true = .025

a = sp.stats.bernoulli.rvs(p=p, size=size, random_state=8228)
b = sp.stats.bernoulli.rvs(p=p*(1+delta_true), size=size, random_state=8228)

print('A - Mean: %.4f, std: %.2f' % (a.mean(), a.std()))
print('B - Mean: %.4f, std: %.2f' % (b.mean(), b.std()))
print('B better A on %.4f = %.4f%%' % (b.mean()- a.mean(), (b.mean()/a.mean()-1)*100))


    

    




A - Mean: 0.0502, std: 0.22
B - Mean: 0.0513, std: 0.22
B better A on 0.0011 = 2.2917%

In [5]:

    

test = ABtest([a, b], BernoulliModel, 5000)

test.test_all()


    

    




ABtest for 2 groups






    
Out[5]:





<abtools.core.abtest.ABtest at 0x7fd5aa8cec50>

In [6]:

    

test.probabilities_df


    

    
Out[6]:







  

    

      

      
group1
      
group2
      
mean
    
  
  

    

      
group1
      
-
      
0.9482
      
0.0501806423946
    
    

      
group2
      
0.0518
      
-
      
0.0513310018794
    
    

      
mean
      
0.0501806423946
      
0.0513310018794
      
-
    
  

In [7]:

    

test.plot()


    

In [ ]: