What is the primary risk of applying Min-Max scaling to a feature containing significant outliers?

Question

Data Science — Hard

What is the primary risk of applying Min-Max scaling to a feature containing significant outliers?

Accepted Answer

Min-Max scaling maps data to a fixed range, typically [0, 1]. If extreme outliers exist, the denominator (max - min) becomes very large, causing the bulk of the data points to be squeezed into an extremely narrow interval near zero. This effectively destroys the variance and discriminative power of the feature for the majority of observations.