Better utilization of correlation between metrics using Principal Component Analysis (PCA)

摘要

Software metrics play an important role in Software Development Life Cycle (SDLC). In this paper we have tried to find the correlation between different software metrics. The research is done using PROMISE data repository of NASA. The positive correlation is found to exist between various software metrics. Large number of metrics available in the software industry raises the need of finding the most significant metrics for better use and control of metrics. For this, the Principal Component Analysis (PCA) based feature selection has been applied on the correlated metrics data set.