Experiments to Parameters and Base Classifiers in the Fitness Function for GA-Ensemble

Abstract

GA-Ensemble is found to be more resistant to outliers and results in simpler predictive models than other ensemble models. The fitness function consists of three parameters (a, b, and p) that limit the number of base classifiers (by b) and control the effects of outliers (by a) to maximize an appropriately chosen p-th percentile of margins. We present the effect of the parameters of a new fitness function as well as the increased complexity of base classifiers to improve predictive accuracy. We use some artificial and real data sets to demonstrate the effect of GA-Ensemble performance at 16 different treatment levels with three different base classifier options and compare to AdaBoost.