plot.FeatureImp() plots the feature importance results of a FeatureImp object.

# S3 method for FeatureImp
plot(x, sort = TRUE, ...)

Arguments

x

A FeatureImp object

sort

logical. Should the features be sorted in descending order? Defaults to TRUE.

...

Further arguments for the objects plot function

Value

ggplot2 plot object

Details

The plot shows the importance per feature.

When n.repetitions in FeatureImp$new was larger than 1, then we get multiple importance estimates per feature. The importance are aggregated and the plot shows the median importance per feature (as dots) and also the 90%-quantile, which helps to understand how much variance the computation has per feature.

See also

Examples

library("rpart")
# We train a tree on the Boston dataset:
data("Boston", package = "MASS")
tree <- rpart(medv ~ ., data = Boston)
y <- Boston$medv
X <- Boston[-which(names(Boston) == "medv")]
mod <- Predictor$new(tree, data = X, y = y)

# Compute feature importances as the performance drop in mean absolute error
imp <- FeatureImp$new(mod, loss = "mae")

# Plot the results directly
plot(imp)