Model Selection with Information Complexity in Multiple Linear Regression Modeling

Abstract

This paper aims to introduce to applied researchers a new family of information model selection criteria in multiple linear regression models. These criteria are known as information complexity (ICOMP) criteria. The paper provides supportive evidence under the R language to show the effectiveness of ICOMP and its tendency to outperform some other traditional criteria: AIC, SBC, etc. This paper also creates a framework on which to base future work in applying ICOMP to more general regression modeling problems in R.