CODEWITHSUNDEEP
pythonscipystatisticskolmogorov-smirnov
MrDinkleburg
MrDinkleburg

Reputation: 109

2 Sample KS test - something seems wrong

I am attempting to run a Kolmogorov-Smirnoff test using the ks_2samp function from scipy to determine if histograms of data are from the same distribution. The returned p-value doesn't seem quite right sometimes though...

For example with this histogram:

Histogram.jpg

aa, bb, cc = ax1.hist(list1, numpy.arange(a-1, b+3, c), alpha = .5, align = 'mid', rwidth=1, linestyle = 'dashed', linewidth = 1.5)

dd, ee, ff = ax1.hist(list2, numpy.arange(a-1, b+3, c), alpha = .5, align = 'mid',rwidth=1)

print ks_2samp(aa, dd)`[1]`

I get a p-value returned of about .96, which really don't seem right...am I doing something wrong? Shouldn't these histograms be different enough that the p-value would be lower?

Upvotes: 1

Views: 881

Answers (1)

cel
cel

Reputation: 31339

ks_2samp applies the Kolmogorov-Smirnov test to two samples and tests the null hypothesis that both come from the same distribution.

Therefore ks_2samp also takes the two samples (here list1 and list2) as input.

ks_2samp(list1, list2)

Upvotes: 3

Related Questions