COCOMO as a schedule prognosis and validation tool: a case study

Abstract

This paper describes a case study in the use of the COCOMO cost estimation model as a tool to provide an independent prognosis and validation of the schedule of a software project at IBM UK Laboratories Ltd, Hursley. Clearly case studies have the danger of being anecdotal however software engineers often work in situations where sufficient historical data is not available to calibrate models to the local environment. It is often necessary for the software engineer to attempt to use such tools on individual projects to justify their further use. This case study describes how we began to use COCOMO and concentrates on some of the problems and benefits which were encountered when trying to use COCOMO in a 'live' development environment. The paper begins by discussing some problems in mapping the COCOMO phases on to the IBM development process. The practical aspects of gathering the development parameters of the model are described and the results of the work are presented in comparison to a schedule assessment using other prognosis techniques and the planned schedule at other milestones in the project's history. Some difficulties experienced in interpreting the data output from the model are discussed. This is followed by a brief comparison with other schedule analysis techniques used in quality assurance. We hope this case study shows that despite the problems in trying to use models such as COCOMO there are significant benefits in helping the user understand what is required to use such tools more effectively to improve software development cost estimates in the future. © 1992 Chapman & Hall.