1

In this post Is there any reason to prefer the AIC or BIC over the other? was given the answer, that

AIC tries to select the model that most adequately describes an unknown

BIC tries to find the TRUE model among the set of candidates.

My question is, how can both this statistics have such different application? Their formulas are "almost" equivalent.

Community
  • 1
Daniel Yefimov
  • 1,222
  • 1
  • 11
  • 27
  • Neither description is accurate. They both penalize the likelihood function for over-parameterizing. They both try to pick a model that fits well. However no model perfectly describes the data. – Michael R. Chernick Mar 30 '17 at 16:11
  • I think the link you provided is a duplicate and so I may elect to close the post for that reason. But at this point I have no votes left for closing. – Michael R. Chernick Mar 30 '17 at 16:15

0 Answers0