This paper reports the results of an empirical investigation of two of the more influential cost estimation techniques: Putnam’s Rayleigh Curve model [1] and Boehm’s COCOMO models [2]. These results indicate weak support for the COCOMO models and no support for the Rayleigh Curve model. They also suggest that neither model should be used indiscriminately across all development environments. On the other hand, the authors did find that it was possible to (1) develop consistent cost models tailored to a particular “class” of software development projects, and (2) improve the precision of these models through the use of actual costs incurred during a project’s early development phases. They conclude by advocating that research in this area should be concerned with developing a more robust understanding of the different “classes” of software projects and associated software development methodologies.
The paper is recommended to those interested in this important topic. The authors clearly describe their study and their findings. Still, the authors could have improved their paper by further exploring their findings in terms of the specific types of software projects included in their data set. Perhaps such an analysis will be forthcoming.