Methodspace - home of the Research Methods community

Please can you help with method for calculating AIC. I have a number of models from a logistic regression and I will like to rank them. Some literature have suggested use of AIC or maximum likelihood. I will like to try the AIC and I have come across a some formulas but am not sure of how to derive its inputs. Or is ther any software that can do it. Hope to hear from you all. txs

Tags: AIC, Akaike, Criteria, Information

Views: 577

Replies to This Discussion

have you looked at:

 

http://en.wikipedia.org/wiki/Akaike_information_criterion ???

 

you basically only need the maximized likelihood (or minimized log-likelihood) and the # of predictors in your logistic regression +1 (for the intercept). i think most software nowadays print those anyways so you dont have to calculate them by hand...

 

using the likelihood operates the same way: the lower the better but information-critertion measures like the Akaike Information Criterion (AIC) or the Bayesian Information Criterion (BIC) help you out by rewarding parsimony and penalizing highly parameterized models or those with many predictors..

 

 

RSS

Follow us:

© 2014   Created by SAGE Publications.

Badges  |  Report an Issue  |  Terms of Service