Dynamic stylometry for defect prediction

Abstract

Allmetrics used by QAs to assess code quality describe the characteristics of a ready-made code. In this article we propose a novel, different approach which tries to catch the dynamics of the coding process. For this purpose we utilize some basic ideas from the science of stylometry. We show how to quantify the process dynamics using several simple metrics.We also present the results of an experiment performed on a group of CS students to validate the prediction power of the proposed model.