Perspectives on Data Science for Software Engineering

One size does not fit all

T. Zimmermann    Microsoft Research, Redmond, WA, United States

Abstract

Not every prediction model can be transferred from one context to another. I've learned this the hard way in my work on defect prediction. The good news is that while prediction models might not always be applicable to other contexts, the methods typically are.

Keywords

Defect prediction models; Cross-project defect prediction

I’ve spent a large part of my career building defect prediction models. The goal of these models is to explain, and in some cases even predict, where defects will occur. Typically such models use a set of metrics, such as lines of codes, complexity, etc. and establish some relationship to the presence or the number of ...