[R] AIC and BIC

Ko-Kang Kevin Wang k.wang at auckland.ac.nz
Mon Apr 26 22:52:38 CEST 2004


----- Original Message ----- 
From: "Talita Leite" <talitaperciano at hotmail.com>
To: <r-help at stat.math.ethz.ch>
Sent: Tuesday, April 27, 2004 12:26 AM
Subject: [R] AIC and BIC

> Hello
> I'm with a doubt using BIC and AIC. I want to know if both of then are a
> to steem the best model to use. How i know which of then to choose?

This is something that can be easily from by asking the dear old google...

But in general, there are various different model selection techniques and
using some sort of Information Criterion is just one of them.  Personally I
tend to go for BIC, as AIC does not take the number of observations into
account.  Having said that when I do model selection, I would do a few
different *IC's just to be sure...



More information about the R-help mailing list