Robust scaling

Robust scaling, also known as robust standardization, is a method of feature scaling that is particularly useful when dealing with datasets containing outliers. Unlike min-max scaling and z-score scaling, which can be sensitive to outliers, robust scaling is designed to be robust in the presence of extreme values. It is especially beneficial when you want to normalize or standardize features while minimizing the impact of extreme values. Robust scaling is also suitable for datasets where the features do not follow a normal distribution and may have skewness or heavy tails.

Here is the formula for robust scaling:

X _ scaled = (X − median) / IQR

Subtracting the median and dividing by the Interquartile Range (IQR)in the scaling process normalizes the data by centering it around the median and scaling it based on the spread represented by the IQR. This normalization helps to mitigate the impact of extreme values, making the scaled values more representative...