Bringing outliers back within acceptable limits

Removing error outliers can be a valid strategy. However, this approach can reduce statistical power, in particular when there are outliers across many variables, because we end up removing big parts of the dataset. An alternative way to handle error outliers is by bringing outliers back within acceptable limits. In practice, what this means is replacing the value of the outliers with some thresholds identified with the IQR proximity rule, the mean and standard deviation, or MAD. In this recipe, we’ll replace outlier values using pandas and feature-engine.

How to do it...

We’ll use the mean and standard deviation to find outliers and then replace their values using pandas and feature-engine:

1. Let’s import the required Python libraries and functions:

   from sklearn.datasets import load_breast_cancer
   from sklearn.model_selection import train_test_split
   from feature_engine.outliers import Winsorizer

2. Load...