machine learning framework
examples

The information about which modules in a software system's future version are potentially defective is a valuable aid for quality managers and testers. Defect prediction promises to indicate these defect-prone modules. In most cases defect prediction models rely on appropriate machine learning methods. Researches at the Software Competence Center Hagenberg are investigating how effective defect prediction models can be constructed in an industrial setting.

In a recent study where a large industrial software system has been analyzed it was investigated what are the appropriate data sources and learning algorithms. It turned out that there is one best data source (static code analysis) and that the fuzzy decision tree learner FS-ID3 from the machine learning framework for Mathematica produced the models which identified the buggy modules with the highest accuracy - and being the most interpretable one!