Module regression 

Regression metrics — error and fit measures over caller-owned prediction slices. Reuses statistics::descriptive for the mean (no re-implementation).

Functions

mae

Mean absolute error.

mse

Mean squared error Σ(yᵢ−ŷᵢ)²/n. None if lengths differ or are empty.

r2_score

Coefficient of determination R² = 1 − SSE/SST. Can be negative for a model worse than the mean. None if lengths differ or SST = 0 (constant target).

rmse

Root mean squared error.