This is a very simple example, I'm simulating normally distributed data in Python, and performing a Kolmogorov - Smirnov Test on my sample to check the goodness-of-fit(of the sample distribution compared with a Normal distribution)


from matplotlib import pyplot as plt

import numpy as np

from scipy import stats

data = stats.norm.rvs(5, 10, size=1000)


print(stats.kstest(data, 'norm'))

enter image description here

KstestResult(statistic=0.57991242997450898, pvalue=0.0)

However, the p-value I get is very small!-> which means I have to reject the Null hypothesis, and conclude that the sample does not follow the distribution.

Could someone explain what I'm doing wrong here?

  • 1
    $\begingroup$ I'm assuming that the KS test will by default compare the sample distribution to a Gaussian distribution with mean 0 and SD 1, which is not the case in your experiment. $\endgroup$ Jan 30, 2018 at 12:23