Dear all, Could anyone please tell me how 'step' or 'stepAIC' works? Does it simply select the model with the smallest AIC from all the possible models? Or does it perform any test eg. whether the decrease in "information content" between a model with a given predictor and another without it is "significant"? Thanks for help! VB
how 'stepAIC' selects?
4 messages · Bókony Veronika, Wuming Gong, Uwe Ligges
B??kony Veronika wrote:
Dear all, Could anyone please tell me how 'step' or 'stepAIC' works? Does it simply select the model with the smallest AIC from all the possible models? Or does it perform any test eg. whether the decrease
No. It works *stepwise*, hence does not inspect all possible models.
in "information content" between a model with a given predictor and another without it is "significant"?
Well, if you believe AIC is the right criteria, an improvement of the AIC is enough to choose another model - no further test will be applied. Uwe Ligges
Thanks for help! VB
______________________________________________ R-help at stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
1 day later
Hi Veronika, StepAIC does not do any test. But you may do a likelihood ratio test on the difference between deviance of two models to judge which model to use, simply like this,
1 - pchisq(deviance.i - deviance.j, df.i - df.j)
And then check whether the returned p value is small enough... Wuming
On 6/18/05, B??kony Veronika <snuggleduck at freemail.hu> wrote:
Dear all, Could anyone please tell me how 'step' or 'stepAIC' works? Does it simply select the model with the smallest AIC from all the possible models? Or does it perform any test eg. whether the decrease in "information content" between a model with a given predictor and another without it is "significant"? Thanks for help! VB
______________________________________________ R-help at stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Wuming Gong wrote:
Hi Veronika, StepAIC does not do any test. But you may do a likelihood ratio test on the difference between deviance of two models to judge which model to use, simply like this,
1 - pchisq(deviance.i - deviance.j, df.i - df.j)
And then check whether the returned p value is small enough...
... but then you are mixing different concepts. Uwe Ligges
Wuming On 6/18/05, B??kony Veronika <snuggleduck at freemail.hu> wrote:
Dear all, Could anyone please tell me how 'step' or 'stepAIC' works? Does it simply select the model with the smallest AIC from all the possible models? Or does it perform any test eg. whether the decrease in "information content" between a model with a given predictor and another without it is "significant"? Thanks for help! VB
______________________________________________ R-help at stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
______________________________________________ R-help at stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html