Permutation Testing for my YouTube video about Hypothesis Testing

Permutation Testing

# Better run this in a Jupyter notebook

import numpy as np
import pandas as pd

green = [32, 34, 38, 28, 32, 34, 38, 28, 33, 50, 32, 39, 29]
red = [33, 32, 39, 29, 33, 32, 39, 29, 33, 8, 32, 39, 29]

green = pd.Series(green)
red = pd.Series(red)

# Get the mean values of the two groups
green.mean(), red.mean() 

# Calculate the means differenece 
original_diff = g.mean() - r.mean()
print(original_diff)

# Permutation Test
diffs = []
refuse_null_hype = []

for i in range(6000):

    s = pd.concat([g,r]).sample(frac=1, replace=False, random_state=i)
    g = s.head(int(0.5 * s.shape[0]))
    r = s.tail(int(0.5 * s.shape[0]))
    
    diff = g.mean() - r.mean()
    diffs.append(diff)
    
    refuse_null_hype.append(diff >= original_diff)

# P Values 
p_value = 0.05
print(
    pd.Series(refuse_numm_hype).mean()
)
print(
   'Can we refuse the null hypothesis?',
    pd.Series(refuse_numm_hype).mean() <= p_value
)
    
# Plot Results
fig, ax = plt.subplots(1, 1, figsize=(14, 8))

pd.Series(diffs).plot(
    title='Diff'
    kind='kde',
    ax=ax
)

ax.axvline(original_diff)

fig.show()