Early-stage product quality prediction by using software process data

To predict projects' quality, cost, and delivery (QCD) in the early stage of software development, Nissin Systems Co. has been implementing risk management, earned value management (EVM), and quality assurance activities. This paper discusses a multiple linear regression analysis by using these process measurement data and derives a relational expression which predicts quality of software products quantitatively. From this analysis, we examine the effects on software products quality of these management factors and quality assurance ones. Based on our results, we have a clear correlation between these activities and software products quality. A prediction of product quality by using process measurement data is shown to be a very effective way to clarify the process factors which affect product quality and to promote the improvement of process factors.