plot.varImp.train: Plotting variable importance measures

In caret: Classification and Regression Training

Plotting variable importance measures

Description

This function produces lattice and ggplot plots of objects with class "varImp.train". More info will be forthcoming.

Usage

## S3 method for class 'varImp.train'
plot(x, top = dim(x$importance)[1], ...)

## S3 method for class 'varImp.train'
ggplot(
  data,
  mapping = NULL,
  top = dim(data$importance)[1],
  ...,
  environment = NULL
)

Arguments

x, data	an object with class varImp.
top	a scalar numeric that specifies the number of variables to be displayed (in order of importance)
...	arguments to pass to the lattice plot function (dotplot and panel.needle)
mapping, environment	unused arguments to make consistent with ggplot2 generic method

Details

For models where there is only one importance value, such a regression models, a "Pareto-type" plot is produced where the variables are ranked by their importance and a needle-plot is used to show the top variables. Horizontal bar charts are used for ggplot.

When there is more than one importance value per predictor, the same plot is produced within conditioning panels for each class. The top predictors are sorted by their average importance.

Value

a lattice plot object

Author(s)

Max Kuhn

caret documentation built on March 31, 2023, 9:49 p.m.