Student research abstract: Threshold analysis of design metrics to detect design flaws

Detection of design flaws at different granularity levels of software can help the software engineer to reduce the testing efforts and maintenance cost. In the context of metric-based analysis, current state of art for the quality assurance tools is to extract the metrics from the source code and analyzed the design complexity. But in case of legacy systems, a software engineer needs to pass through the re-engineering process. In this study, I propose a methodology to investigate the threshold effect of software design metrics in order to detect design flaws and its effect over the granularity level of software. Moreover, I will use some statistical methods and machine learning techniques to derive and validate the effect of thresholds over the NASA and open source datasets retrieve from the PROMISE repository.