| During the last few years, several package-level metrics have been developed and used to characterize the attributes of packages in object-oriented software design. These metrics provide ways to evaluate the quality of software. But how do we know which metrics are useful in capturing important quality attributes. Empirical studies are needed to provide relevant answers.;The proposed work will study sets of package-level metrics and empirically validate them against some implementation and quality assessment attributes. In addition, these metrics have been implemented in this work, to automate the extraction of the package level metrics. The collected metrics have been investigated to explore their capability to identify and predict pre- and post-release fault, change density and implementation effort for software packages. The results show that some of the package-level metrics are good indicators for change prediction, fault prediction and implementation effort estimation, even after controlling for package size. |
