A benchmark-based approach for ranking root causes of performance problems in software development

Abstract

In previous work we proposed a performance analysis model for automatically identifying potential root causes of performance problems in personal software development. In this paper we present an approach for automatically ranking those potential root causes based on a cost-benefit estimate that takes into account historical data. The approach was applied for the Personal Software Process, taking advantage of a large data set referring to more than 30, 000 projects, but can be replicated in other contexts.