ggRandomForests (version 3.1.2)

plot.gg_vimp: Plot a gg_vimp object, extracted variable importance of a rfsrc object

Description

Draws a horizontal bar chart of the VIMP scores extracted by gg_vimp. Each bar represents one predictor; bar length is proportional to its permutation VIMP -- the average rise in OOB prediction error when that predictor's OOB values are randomly shuffled. Predictors are sorted in descending order of importance so the most influential variables appear at the top.

Usage

# S3 method for gg_vimp
plot(x, relative, lbls, ...)

Value

ggplot object

Arguments

x

gg_vimp object created from a rfsrc object

relative

should we plot vimp or relative vimp. Defaults to vimp.

lbls

A vector of alternative variable labels. Item names should be the same as the variable names.

...

optional arguments passed to gg_vimp if necessary

Details

Bars are coloured by the positive flag: a bar at or below zero (non-positive VIMP) is colour-coded differently to flag predictors that hurt OOB accuracy when their signal is removed -- usually a sign of collinearity or a very noisy variable. In a well-behaved forest most bars are positive; the colour distinction matters when a handful are not.

References

Breiman L. (2001). Random forests, Machine Learning, 45:5-32.

Ishwaran H. and Kogalur U.B. (2007). Random survival forests for R, Rnews, 7(2):25-31.

Ishwaran H. and Kogalur U.B. randomForestSRC: Random Forests for Survival, Regression and Classification. R package version >= 3.4.0. https://cran.r-project.org/package=randomForestSRC

See Also

gg_vimp

Examples

## ------------------------------------------------------------
## classification example
## ------------------------------------------------------------
## -------- iris data
rfsrc_iris <- randomForestSRC::rfsrc(Species ~ ., data = iris)
gg_dta <- gg_vimp(rfsrc_iris)
plot(gg_dta)

## ------------------------------------------------------------
## regression example
## ------------------------------------------------------------
## -------- air quality data
rfsrc_airq <- randomForestSRC::rfsrc(Ozone ~ ., airquality)
gg_dta <- gg_vimp(rfsrc_airq)
plot(gg_dta)