Effort estimation of use cases for incremental large-scale software development

This paper describes an industrial study of an effort estimation method based on use cases, the Use Case Points method. The original method was adapted to incremental development and evaluated on a large industrial system with modification of software from the previous release. We modified the following elements of the original method: a) complexity assessment of actors and use cases, and b) the handling of non-functional requirements and team factors that may affect effort. For incremental development, we added two elements to the method: c) counting both all and the modified actors and transactions of use cases, and d) effort estimation for secondary changes of software not reflected in use cases. We finally extended the method to: e) cover all development effort in a very large project. The method was calibrated using data from one release and it produced an estimate for the successive release that was only 17% lower than the actual effort. The study identified factors affecting effort on large projects with incremental development. It also showed how these factors can be calibrated for a specific context and produce relatively accurate estimates.