projection_importance: Projection-coefficient variable importance.

In ppforest2: Projection Pursuit Oblique Decision Trees and Random Forests

Projection-coefficient variable importance.

Description

The projection-based importance (VI2): each split's scaled absolute projection coefficients ('|a_j| * sigma_j') aggregated into a per-feature score, averaged over the non-degenerate trees of a forest. Unlike permuted_importance and weighted_importance, this measure is not OOB-based — it depends only on the fitted projector geometry, is computed eagerly at fit time (cheap), and is available for both single trees (pptr) and forests (pprf).

Usage

projection_importance(model)

Arguments

model

A pptr or pprf model.

Details

**Sign semantics.** Entries are non-negative by construction (absolute coefficients scaled by each feature's standard deviation). Rely on the ranking rather than re-normalizing.

Value

A non-negative numeric vector, one entry per feature.

See Also

permuted_importance, weighted_importance

ppforest2 documentation built on July 21, 2026, 9:07 a.m.