vimp: Variable Importance

In PhilippPro/vimp: Variable Importance

Description

With the help of this function the permutation variable importance for random forests can be created for any measure that is available in the mlr package.

Usage

vimp(mod, measures = list(auc), task, data)

Arguments

mod	An object of class randomForest, as that created by the function randomForest with option keep.inbag = TRUE
measures	List of performance measure(s) of mlr to evaluate. Default is auc only.
task	Learning task created by the function makeClassifTask or makeRegrTask of mlr.

Value

Returns a dataframe with a column for each desired measure.

Examples

library(mlr)
library(randomForest)

# Classification
data = getTaskData(iris.task)
mod = randomForest(Species ~., data = data, ntree = 100, keep.inbag = TRUE)
results = vimp(mod, measures = list(mmce, multiclass.au1u, multiclass.brier), task = iris.task, data = data)

PhilippPro/vimp documentation built on May 8, 2019, 2:34 a.m.