An analysis of the aggregated coupling’s suitability for software defect prediction

Data sets for the ant (1.7), jEdit (3.2, 4.0, 4.1, 4.2, 4.3) tomcat (6.0) open source software systems containing the following features:- 36 conceptual coupling based features- 9 relevant properties based features- 36 aggregated coupling based features- values for CBO and RFC metrics- class label - 1 for defective, 0 for non-defective
It also contains a file with the experimental evaluation results, on different subsets of these metrics using different machine learning models from the Orange framework.