Implications

Adding visualization here showing stratification:

Example: Comparing the Click-Through Rates(CTR) for A page and B page.

• Analysis without stratification:

A performs a little better than B

• **Analysis on removal of over sampling for Old people in the data set.

B performs better than A

• Analysis when we stratify on one of the features("Age"):

We see that the performance of A and B depends on the age of the person and that is the reason we were getting flawed inferences earlier.

****

Go to the Notebook for reference!!!