A Study of New Variable Selection Method Within a Framework of Real-Coded Genetic Algorithm

Recently, variable selection and parameter optimization are becoming increasingly important. Regarding parameter optimization, real-coded genetic algorithms (RCGA) have received attention due to their strong searching ability and flexibility. The Akaike information criterion (AIC) or Bayesian information criterion (BIC) are traditionally used as variable selection criteria. These criteria estimate the relative quality of analysis models for a given set of data, but they cannot be used to evaluate the importance of the variables themselves. This paper proposes a new variable selection method that applies RCGA. This new variable selection method contains two primary components. One is a new variable selection criterion, and the other is a method for estimating the progress of RCGA optimization. The effectiveness of this new variable selection method is confirmed through application to the sum of squares function, which is a nonlinear test function.