A hybridized consistent Akaike type information criterion for regression models in the presence of multicollinearity

Dünder E.

COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION, 2023 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Publication Date: 2023
  • Doi Number: 10.1080/03610918.2023.2169710
  • Journal Name: COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Applied Science & Technology Source, Business Source Elite, Business Source Premier, CAB Abstracts, Compendex, Computer & Applied Sciences, Veterinary Science Database, zbMATH, Civil Engineering Abstracts
  • Keywords: Information criteria, Model selection, Regression modeling, SELECTION, COMPLEXITY
  • Ondokuz Mayıs University Affiliated: Yes

Abstract

Consistent Akaike information criterion (CAIC) is an adjusted form of classical AIC. This criterion was developed by modifying the penalty. As a result, we propose a novel AIC type criterion, called CAIC (n alpha). The proposed criterion includes a dynamic parameter for controlling the penalty further. The distinctive feature of CAIC (n alpha) is to penalize multicollinearity level considering the information complexity measures. CAIC (n alpha) requires the alpha parameter, and in addition, a procedure is proposed to estimate alpha based on the information complexity of the regression model. Monte Carlo simulations and real data set examples demonstrate that CAIC (n alpha) performs better than classical information criteria for the potential multicollinearity problems.