### logistic regression - low-birthweight data from VR library(MASS) data(birthwt) ### same pre-processing as VR attach(birthwt) race <- factor(race, labels=c("white", "black", "other")) ptd <- factor(ptl>0) ftv <- factor(ftv); levels(ftv)[-(1:2)] <- "2+" bwt <- data.frame(low=factor(low), age, lwt, race, smoke=(smoke>0), ptd, ht=(ht>0), ui=(ui>0), ftv) detach(); rm(race, ptd, ftv) ### stepwise AIC selection, as in textbook fit0 <- glm(low ~ ., family=binomial, data=bwt) fit1 <- stepAIC(fit0, ~., data=bwt, trace=F) fit2 <- stepAIC(fit0, ~.^2 + I(scale(age)^2) + I(scale(lwt)^2) ,data=bwt, trace=F) fit3 <- stepAIC(fit1, ~.^2 + I(scale(age)^2) + I(scale(lwt)^2), data=bwt, trace=F) fit1a <- stepAIC(fit0, trace=F, k=log(nrow(bwt)), data=bwt) fit2a <- stepAIC(fit0, ~.^2 + I(scale(age)^2) + I(scale(lwt)^2), trace=F, k=log(nrow(bwt)),data=bwt) fit3a <- stepAIC(fit1a, ~.^2 + I(scale(age)^2) + I(scale(lwt)^2), trace=F, k=log(nrow(bwt)),data=bwt)