Empirical validation of object-oriented metrics on NASA for fault prediction

Abstract

In this study, empirical analysis is carried out to validate Chidamber and Kemerer metrics suite for predicting fault proneness when taking fault severity levels into account. The results, based on KC1 data set from the public NASA repository, indicate that 1) WMC, NOC, LCOM and CBO are reliable metrics for fault-proneness of classes across fault severity in the prediction models built by decision tree method, and 2) only CBO is reliable metrics for fault-proneness of classes across fault severity in the prediction models built by binary logistic regression method. © 2011 Springer-Verlag.