Calculate Importance using Johnson's Relative Weights

This function is a more efficient, though less precise, alternative to kda_importance_domir(). While kda_importance_domir() can become computationally intensive with more than ~15 predictors, this function scales better to larger models.

For linear regression models, relative importance is computed from each predictor's contribution to model $(R^2)$. For logistic regression models, relative importance is computed from each predictor's contribution to pseudo-$(R^2)$.

Usage

kda_importance_jrw(model)

Arguments

model

A model object.

Value

A list containing:

• out: A tibble with predictor importance metrics (raw, ratio, percent, and rank).

• jrw: The Johnson's Relative Weights object computed using rwa::rwa().

See also

kda_importance_domir(), rwa::rwa()