Validation of measurement tools to extract metrics from open source projects

Abstract

Software measurement can play a major role in ensuring the quality and reliability of software products. The measurement activities require appropriate tools to collect relevant metric data. Currently, there are several such tools available for software measurement. The main objective of this paper is to discuss the validation of results when using a combination of multiple measurement tools especially for products built using object-oriented techniques and languages. In this paper, we highlight four tools for collecting metric data, in our case from several Java-based open source projects. Our findings show that the tools provide different results for similar metrics and we discuss the methods used to validate the results reported by the tools. Our research is currently based on the work of Card and Glass, who argue that design complexity measures (data complexity and structural complexity) are indicators/predictors of procedural/cyclomatic complexity (decision counts) and errors (discovered from system tests).