RF Variable Importance for Arbitrary Measures

Computes the random forest variable importance (VIMP) for the conditional inference random forest (cforest) of the 'party' package. Includes a function (varImp) that computes the VIMP for arbitrary measures from the 'measures' package. For calculating the VIMP regarding the measures accuracy and AUC two extra functions exist (varImpACC and varImpAUC).

varImp

Random forest variable importance for arbitrary measures of the measures package, which contains the biggest collection of measures for regression and classification in R.

Installation

The development version

devtools::install_github("mlr-org/measures")
devtools::install_github("PhilippPro/varImp")
iris.cf <- cforest(Species ~ ., data = iris, control = cforest_unbiased(mtry = 2, ntree = 50))
varImp(object = iris.cf, measure = "multiclass.Brier")
varImpACC(object = iris.cf)
# Two classes:
iris2 = iris
iris2$Species = factor(iris$Species == "versicolor")
iris.cf = cforest(Species ~ ., data = iris2,control = cforest_unbiased(mtry = 2, ntree = 50))
varImpAUC(object = iris.cf)