Abstract:
Maintenance costs are a big concern for the software industry and users. Maintainability depends on the changeability of the software. One way of measuring changeability is to measure coupling between different modules of systems. High coupling in software is avoided because that makes system more rigid and hard to change. The first goal of this research is to empirically explore the effect of coupling on changes. To measure the effect of coupling on changes, we examined coupling and changes in ten open source Java systems. Extensive research is on going to develop software metrics and tool to measure the quality of software. Various metrics have been proposed to measure external software quality attributes like changeability, ripple effects of changes, impact analysis, etc. The second goal is to empirically validate the coverage of coupling between objects (CBO) metric proposed by Chidamber and Kemerer [5] and propose a new metric new dependency metric NDM to measure the couplings that are overlooked by the CBO metric. We used spearman correlation coefficient statistic for results validation and found that NDM measures more coupling between classes than the CBO metric and that high coupling is negatively related with changes.