GIC {MASSExtra}

Intermediate Information Criterion

Description

An AIC-variant criterion that weights complexity with a penalty mid-way between 2 (as for AIC) and log(n) (as for BIC). I.e. "not too soft" and "not too hard", just "Glodilocks".

Usage

GIC(object)

Arguments

object

a fitted model object for which the criterion is desired

Value

The GIC criterion value

Examples

gm <- glm.nb(Days ~ Sex/(Age + Eth*Lrn), quine)
c(AIC = AIC(gm), GIC = GIC(gm), BIC = BIC(gm))