bestModel_sigma: Selecting the best model, by bootstrap and Cross Validation,...

In cdv04/ACTR:

Description

6 linear models are compared using Cross Validation on the basis of the least error of prediction

Usage

bestModel_sigma(errMod_Sigma.list)

Arguments

errMod_Sigma

A list containing the error of prediction for each model and each bootstrap samples

Details

The 6 linear models are :

mod1 : wSigmaC_lg ~ 1

mod2 : wSigmaC_lg ~ wSigmaA_lg

mod3 : wMuC_lg ~ wMuA_lg + wSigmaA_lg

mod4 : wSigmaC_lg ~ wSigmaA_lg + I(wSigmaA_lg ^2)

mod5 : wSigmaC_lg ~ wSigmaA_lg + I(wSigmaA_lg ^2) + wMuA_lg

mod6 : wSigmaC_lg ~ wSigmaA_lg + I(wSigmaA_lg ^2) + wMuA_lg + I(wMuA_lg ^2)

Value

The function return the formula of the best selected model

Examples

data(cipr)
ciprKP <- subset_data(cipr, Class,6)
PAC.bt <- lapply(ciprKP.bt,est_PAC,Class)
errMod_mu <- lapply(PAC.bt,msep_mu)
errMod_sigma <- lapply(PAC.bt,msep_sigma)
bm_sigma <- bestModel_mu(errMod_sigma)

cdv04/ACTR documentation built on May 13, 2019, 2:42 p.m.