COCOMO-U: An Extension of COCOMO II for Cost Estimation with Uncertainty

It is well documented that the software industry suffers from frequent cost overruns, and the software cost estimation remains a challenging issue. A contributing factor is, we believe, the inherent uncertainty of assessment of cost. Considering the uncertainty with cost drivers and representing the cost as a distribution of values can help us better understand the uncertainty of cost estimations and provide decision support for budge setting or cost control. In this paper, we use Bayesian belief networks to extend the COCOMO II for cost estimation with uncertainty, and construct the probabilistic cost model COCOMO-U. This model can be used to deal with the uncertainties of cost factors and estimate the cost probability distribution. We also demonstrate how the COCOMO-U is used to provide decision support for software development budget setting and cost control in a case study.