Potential of artificial intelligence based feature selection methods in regression models

Pattern recognition based on learning approaches is regarded as one of the disciplines of AI. Floating search methods, developed originally for feature selection problems in statistical pattern recognition, are applicable to a much wider class of problems outside pattern recognition. They have the potential to find an optimal subset of variables maximizing any criterion adopted for the problem at hand-eliminating the so-called nesting effect from which traditional algorithms suffer. One such application area is multiple regression, where floating search methods represent a computationally feasible alternative to classical methods for finding the optimal set of regressors.