warpingUnder: warpingUnder function

In dalpozz/warping: Probability warping and calibration with undersampling

Description

Function used to asses the wrapping effect of undersampling on the posterior probability. use a large number of folds to make sure we use almost all the dataset for training

Usage

warpingUnder(formula, data, algo, task.id = "cv", positive = 1,
  costs = NULL, nCV = 10, B = 1, nFold = 100, ncore = 1,
  dirPlot = NA, verbose = TRUE, ...)

Arguments

formula	A formula of the form y ~ x1 + x2 + ...
data	Data frame from which variables specified in formula are preferentially to be taken
algo	classification algorithm supported by mlr package
task.id	name of the task
positive	value of the positive (minority) class
costs	cost matrix
nCV	number of repetation of the Cross Validation (CV)
B	number of models to create for each fold of the CV
nFold	number of folds of the CV
ncore	number of cores to use in multicore computation
dirPlot	directory where to save plots (set dirPlot=NA to avoid plots)
verbose	print extra information (logical variable)
...	extra parameters passed to mlr train function

Value

The function returns a list:

results	results of undersampling
probs	probabilities
mres	mres
meltResALL	meltResALL

Examples

library(mlbench)
data(Ionosphere)
library(warping)
res <- warpingUnder(Class ~., Ionosphere, "randomForest", task.id="rf_Ionosphere", positive="bad", nCV=3, B=1, nFold=5)

dalpozz/warping documentation built on May 14, 2019, 3:32 p.m.