Plot a GAM's fitted smooths, with fancyfx

The partial effect of each smooth term on the log-odds scale, with its standard error band and a rug showing where the data actually is. This is the exact version of partial_effects() for a GAM: read out of the fitted model rather than reconstructed by prediction, so it carries uncertainty, which a partial dependence curve cannot.

Usage

plot_gam_smooths(model, vars = NULL, path = NULL)

Arguments

model: a fitted model from fit_patch_model()
vars: which smooths to draw; NULL uses all of them
path: optional file to write the plot to instead of returning it

Value

a patchwork/ggplot object, or path invisibly when writing

Details

Drawn by fancyfx, which is a Suggests — a run without it still gets the generic partial effect curves.

Why the axes read in standard deviations

The smooths belong to the model, and the model was fitted on the recipe's output — so x is whatever the recipe made of the predictor. With the default covariates.normalize: true that is standard deviations from the mean, and the rug is taken from the same baked data so the two line up. Set covariates.normalize: false to have these read in the covariate's own units; it costs a tree model nothing, and a GAM little.

Examples