An Empirical Validation of Coupling Metrics Using Automated Refactoring

The validation of software metrics has received much interest over the years due to a desire to promote metrics which are both well-founded theoretically and have also been shown empirically to reflect our intuition and be practically useful. In this paper we describe how we used automatic refactoring to investigate the changes which occur to a number of metrics as software evolves. We use the concept of software volatility to quantify our comparison of the metrics.