Classification of software components based on clustering

This thesis demonstrates how in different phases of the software life cycle, software components that have similar software metrics can be grouped into homogeneous clusters. We use multi-variate analysis techniques to group similar software components. The results were applied on several real case studies from NASA and open source software. We obtained process and product related metrics during the requirements specification, product related metrics at the architectural level and code metrics from operational stage for several case studies. We implemented clustering analysis using these metrics and validated the results. This analysis makes it possible to rank the clusters and assign similar development and validation tasks for all the components in a cluster, as the components in a cluster have similar metrics and hence tend to behave alike