Profiling, extracting, and analyzing dynamic software metrics

This thesis presents a methodology for the analysis of software executions aimed at profiling software, extracting dynamic software metrics, and then analyzing those metrics with the goal of assisting software quality researchers. The methodology is implemented in a toolkit which consists of an event-based profiler which collects more accurate data than existing profilers, and a program called MetricView that derives and extracts dynamic metrics from the generated profiles. The toolkit was designed to be modular and flexible, allowing analysts and developers to easily extend its functionality to derive new or custom dynamic software metrics. We demonstrate the effectiveness and usefulness of DynaMEAT by applying it to several open-source projects of varying sizes