Akaike Information Criterion

The Akaike Information Criterion (AIC) is one of the first and among the most widely used model selection criteria. In general, a model selection criterion is a measure that summarizes how effectively a statistical model balances the competing objectives of parsimony and fidelity to the data used in its construction. AIC can be used to rank order a defined collection of candidate models and to identify a “best” model from among these candidates.

AIC was first introduced by Hirotugu Akaike in 1973 as an extension to the maximum likelihood principle, which assumes that the size and structure of a statistical model are known and that the data need only be used to estimate the associated (unknown) model parameters. Akaike described AIC as being based on an ...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles