EVALUATING THE CMARS PERFORMANCE FOR MODELING NONLINEARITIES

Multivariate Adaptive Regression Splines (MARS) is a very popular nonparametric regression method particularly useful for modeling nonlinear relationships that may exist among the variables. Recently, we developed CMARS method as an alternative to backward stepwise part of the MARS algorithm. Comparative studies have indicated that CMARS performs better than MARS for modeling nonlinear relationships. In those studies, however, only main and two-factor interaction effects were sufficient to model the nonlinearity between the variables in the data sets. In this study, therefore, we aim at evaluating the model performances when there is a need for representing higher-order interaction effects in a nonlinear model. Results based on the comparison studies show that CMARS method performs better than MARS method according to most of the performance measures.