Why might QuantileTransformer improve robustness but change a linear model's interpretation?
Classic MLLevel: IntermediateRank transformation
Why might QuantileTransformer improve robustness but change a linear model's interpretation?
Hints
- Think ranks rather than original units.
- Nonlinear maps need not preserve Pearson correlation.
Answer
It maps empirical quantiles nonlinearly, compressing extremes. Order is largely preserved, but distances, correlations and linear relationships can change.
Source and licence: scikit-learn 1.7.2 — preprocessing · BSD-3-Clause · answer: Quantoros editorial