Comparison of conventional approaches and Soft-Computing approaches for Software Quality Prediction

Managing sofrware development and maintenance projects requires early knowledge about quality and effort needed for achieving a necessaly quality Level. QuaLio prediction models can identify outlying software components thaf might cause potential quality problems. Quality prediction is based on experience with similar predecessor projects constructing a relationship between the output usually the number of errors and some hind of (input here we use complexity metrics to the quality of a sofhvare development project. Two approaches are presented to build quality prediction models : Mullilinear discriminant analysis as one example for conventional approaches and Fuzzy Expert-Systems generated by Genetic Algorithms. Using the capabiliol of Genetic Algorithms, the fuzzy rules can be automatically generated from example datu to reduce the cost and improve the accuracy. The generated quality ,model with respect to changes provides both quality of j i t (according to past data) ana' predictive accuracy (according to ongoing projects). The comparison of the appoaches gives an answer on the effectiveness and the eficiericy of a SofirComputing approach. effort expended on correcting the sources of defect. The approach to improve software development productivity is therefore based on the improvement of the quality, especially the correctness and the maintainability. In order to achieve an indication of software quality, the software must be subjected to measurement. This is accomplished through the use of metrics. The evaluation is done, based on statistical techniques that relate specific quantified product requirements to some attributes of quality. This presentation introduces fuzzy expert system techniques as a basis for constructing quality based productivity prediction models that can identify outlying software components that might cause potential quality problems, thus requesting additional effort. This effort extremely shows up in maintenance projects done years later by new developers. The presentation is organized as follows. The second section presents a brief overview of the background and the problems associated with metricbased decision models (e.g. vagueness in reasoning). The following section discusses the fuzzy sets, their application in fuzzy classification and the construction of a fuzzy classification system for software quality control. The next section describes the extraction of a Fuzzy Expert System out of project data using Genetic Algorithms. Finally, experimental results are provided to demonstrate the effectiveness of the approach in the area of errorand change-prediction. 2. Fuzzy Data Analysis