Given the complexity and sophistication of many contemporary software systems, it is often difficult to gauge the effectiveness, maintainability, extensibility, and efficiency of their underlying software components. A strategy to evaluate the qualitative attributes of a system's components is to use software metrics as quantitative predictors. We present a fusion strategy that combines the predicted qualitative assessments from multiple classifiers with the anticipated outcome that the aggregated predictions are superior to any individual classifier prediction. Multiple linear classifiers are presented with different, randomly selected, subsets of software metrics. In this study, the software components are from a sophisticated biomedical data analysis system, while the external reference test is a thorough assessment of both complexity and maintainability, by a software architect, of each system component. The fuzzy integration results are compared against the best individual classifier operating on a software metric subset
Discussion(0)
No comments yet. Be the first to comment.