The analysis of software metrics for design complexity and its impact on reusability

Reusability and complexity of software are key features of software quality. Such design metrics are considered to have potential for improvement of software quality and developer productivity. This study performs analysis of Object-Oriented design complexity metrics and its relation with reusability. To be precise the paper considers the most cited Chidamber and Kemerer (CK) metric suite. This study provides empirical evidence in support of the role of design metrics specially CK metrics in estimating reusability of software components. In this paper the competence and effectiveness of machine learning regression techniques are also examined. An experiment is performed to analyze comparative study of Multi linear regression, Model Tree M5P, Meta-learning scheme Additive Regression and Isotonic Regression. This experiment is performed by using data values from projects existing in real world. The results indicate that the complexity is considerably associated with reusability of software. For this study the paper uses Weka tool. The paper believes that the results from this study provide significant suggestions for designing high quality software applications using Object-Oriented and Component-Based approach and identifies the better regression algorithm for reusability estimation using complexity metrics.