Calibrating Software Cost Models to Department of Defense Databases: A Review of Ten Studies - There are many sophisticated models and methods for estimating the size, cost, and schedule of software projects. However, the ability to accurately estimate software cost, size, or schedule is still dubious. In general, the predictive accuracy of models for estimating software development cost and schedule has not been shown to be better than “within 25 percent of actual cost or schedule, about one half of the time”, especially for Department of Defense software efforts. The same is true for software size estimating models and methods, although there are some studies that have shown superior results. For software support (or maintenance) cost estimation, no model has been demonstrated to be accurate. This paper presents a summary of efforts performed to date which demonstrate the accuracy (or lack thereof) of software models. The results of several studies in the areas of software development cost and schedule estimation, size estimation, and support cost estimation are presented to show what these models can and cannot do. Some ideas for improvement are also presented, including the results of some studies which may lead to a resolution of the accuracy conundrum which currently exists...
Software Development Cost Estimation Approaches -- A Survey - This paper by Dr. Barry Boehm, summarizes several classes of software cost estimation models and techniques: parametric models, expertise-based techniques, learning-oriented techniques, dynamics-based models, regression-based models, and composite-Bayesian techniques for integrating expertise-based and regression-based models. Experience to date indicates that neural-net and dynamics-based techniques are less mature than the other classes of techniques, but that all classes of techniques are challenged by the rapid pace of change in software technology. This link downloads a PDF file.
DACS Gold Practice Initiative
