kfold.idx: Indices for K-fold validation

In LGDtoolkit: Collection of Tools for LGD Rating Model Development

Indices for K-fold validation

Description

kfold.idx provides indices for K-fold validation.

Usage

kfold.idx(target, k = 10, type, num.strata = 4, seed = 2191)

Arguments

target	Continuous target variable.
k	Number of folds. If k is equal or greater than the number of observations of target variable, then validation procedure is equivalent to leave one out cross-validation (LOOCV) method. Default is set to 10.
type	Sampling type. Possible options are "random" and "stratified".
num.strata	Number of strata for "stratified" type. Default is 4.
seed	Random seed needed for ensuring the result reproducibility. Default is 2191.

Value

The command kfold.idx returns a list of k folds estimation and validation indices.

Examples

library(monobin)
library(LGDtoolkit)
data(lgd.ds.c)
#random k-folds
kf.r <- LGDtoolkit::kfold.idx(target = lgd.ds.c$lgd, k = 5, 
			type = "random", seed = 2211)
sapply(kf.r, function(x) c(mean(lgd.ds.c$lgd[x[[1]]]), mean(lgd.ds.c$lgd[x[[2]]])))
sapply(kf.r, function(x) length(x[[2]]))
#stratified k-folds
kf.s <- LGDtoolkit::kfold.idx(target = lgd.ds.c$lgd, k = 5, 
                              type = "stratified", num.strata = 10, seed = 2211)
sapply(kf.s, function(x) c(mean(lgd.ds.c$lgd[x[[1]]]), mean(lgd.ds.c$lgd[x[[2]]])))
sapply(kf.s, function(x) length(x[[2]]))

LGDtoolkit documentation built on May 31, 2023, 9:11 p.m.