Software project risk estimation using bayesian belief networks

Abstract

Software development projects offer unique challenges and risks because of their abstract and diverse nature. Risk Analysis and management in software engineering is an emerging part of project management. Software project risk estimation has been a problem in software engineering for many years. A substantial amount of literature has been written to address this issue but this literature does not fully address the concern of the problem. A number of risk estimation models and methods are proposed, based on expert judgment and estimation by analogy, but this area still remains an issue. Normally only expert judgment has been used to address this issue but it does not address the concern of the problem fully because of the diverse nature of risk factors/drivers. We propose the use of BBNs (Bayesian Belief Networks), which are quite famous for their applicability in software engineering areas, to support expert judgment in software risk estimation. Approach for identifying risks is usually separate from estimating those identified risks. We integrate project overspend, schedule overrun, required functionality missing, and poor quality of product with the quantitative assessment of probability of risks.