|Authors||B. C. D. Anda, H. C. Benestad and S. E. Hove|
|Title||A Multiple-Case Study of Effort Estimation Based on Use Case Points|
|Afilliation||Software Engineering, Software Engineering|
|Publication Type||Proceedings, refereed|
|Year of Publication||2005|
|Conference Name||ISESE'2005 (Fourth International Symposium on Empirical Software Engineering)|
|Publisher||IEEE Computer Society, Noosa, Australia, November 17-18|
There is much interest in industry for the estimation of software development effort based on use cases, but little scientific evaluation of applying use cases in estimation has been reported. We investigated one particular method, the use case points method, in a multiple-case study. The Software Engineering Department at Simula issued a tender for a system, and 35 companies responded, with estimates ranging from 78 to 654 hours. We chose four companies to develop the system. They all implemented the same functionality, but their development processes varied, ranging from a light, mainly code-and-fix process, to a heavy process with much emphasis on analysis and design. The use case points method estimated this project to 430 hours. This was equal to the actual effort spent on implementing the system by the company with the lightest development process. In our opinion, the results from this study may represent a basis for measuring size of the use cases and choosing productivity factor (hours per use case point) when estimating based on use cases. The three other companies spent 587, 829 and 943 hours respectively, showing that a heavier development process and more emphasis on non-functional requirements may increase effort by more than 100%.