The impact of measurement scale and correlation structure on classification performance of inductive learning and statistical methods

摘要

This is a comparative study of inductive learning and statistical methods using the simulation approach to provide a generalizable results. The purpose of this study is to investigate the impact of measurement scale of explanatory variables on the relative performance of the statistical method (probit) and the inductive learning method (ID3) and to examine the impact of correlation structure on the classification behavior of the probit method and the ID3 method. The simulation results show that the relative classification accuracy of ID3 to probit increases as the proportion of binary variables increases in the classification model, and that the relative accuracy of ID3 to probit is higher when the covariance matrices are unequal among populations than when the covariance matrices are equal among populations. The empirical tests on ID3 reveal that the classification accuracy of ID3 is lower when the covariance matrices are unequal among populations than when the covariance matrices are equal among populations and that the classification accuracy of ID3 decreases as the correlations among explanatory variables increases.