step_down {MASSExtra}

Naive backeward elimination

Description

A simple facility to refine models by backward elimination. Covers cases where drop_term works but step_AIC does not

Usage

step_down(object, ..., trace = FALSE, k)

Arguments

object

A fitted model object

...

additional arguments passed to drop_term such as k

trace

logical: do you want a trace of the process printed?

k

penalty (default 2, as for AIC)

Value

A refined fitted model object

Examples

fm <- lm(medv ~ . + (rm + tax + lstat)^2 +
           I((rm - 6)^2) + I((tax - 400)^2) + I((lstat - 12)^2), Boston)
sfm <- step_down(fm, trace = TRUE, k = "bic")