Combining object‐oriented metrics and centrality measures to predict faults in object‐oriented software: An empirical validation

Many object-oriented metrics have been proposed in the literature to measure various structural properties of software. Furthermore, many centrality measures introduced identify central nodes large networks. However, few studies used them dependencies software systems. In fact, measures, as opposed most traditional that mainly focus on intrinsic classes, can be better model control flow and important classes a system. This paper aims (1) investigate relationships between (2) explore ability their combination support fault-proneness prediction from different perspectives (fault-prone fault severity, number faults). addressed fault-prone perspectives, using metrics. The main motivation here is fact if information captured by related proneness complementary improves performance significantly. We size, complexity, coupling addition measures. collected data 20 versions five open-source Java first studied selected proneness. Then, we built models predict several machine learning algorithms. addition, class contains high severity fault, faults class. Results indicate with well has no significant impact, according collected, quality severity. Moreover, also both cross-version cross-system validation.