Python Feature Engineering Cookbook

There's more...

We can winsorize many variables at the time by utilizing the open source package, Feature-engine. To do this, we need to load the libraries and the data as we did in step 1 and step 2 of the recipe in the How to do it... section. Next, we need to import Feature-engine:

1. Import Winsorizer from Feature-engine:

from feature_engine.outlier_removers import Winsorizer

2. Set up a Feature-engine Winsorizer indicating which variables we want to winsorize:

windsorizer = Winsorizer(distribution='quantiles', tail='both', variables=['RM', 'LSTAT', 'CRIM'])

With Winsorizer from Feature-engine, we can replace the values by the percentiles at the left and right tails or only at one of the tails by setting the argument tail to either both ...