Bayesian analysis of software cost and quality models

Abstract

Due to the pervasive nature of software, software-engineering practitioners have continuously expressed their concerns over their inability to accurately predict the cost, schedule and quality of a software product under development. Thus, one of the most important objectives of the software engineering community has been to develop useful models that constructively explain the software development lifecycle and accurately predict the cost, schedule and quality of developing a software product. Most of the existing parametric models have been empirically calibrated to actual data from completed software projects. The most commonly used technique for empirical calibration has been the popular classical multiple regression approach. This approach imposes a few restrictions often violated by software engineering data and has resulted in the development of inaccurate empirical models that do not perform very well. The focus of this dissertation is to explain the drawbacks of the multiple regression approach for software engineering data and discuss the Bayesian approach which alleviates a few of the problems faced by the multiple regression approach.