hard20 pts · first answer countsmodel-monitoringstatistical-tests

Every feature fails the test

Your daily drift job runs a KS test on 1 crore rows per feature. Nearly every feature now has p < 0.001, yet the histograms look almost identical. Why?

A.The KS test is invalid for more than 1 lakh rowsB.p-values below 0.001 mean the job has a bugC.KS needs normal data, so the p-values are randomD.With huge samples, tests detect tiny, unimportant differences

Discussion 0

Join free to post in the discussion.

Every feature fails the test

Discussion 0

Related questions