Analyzing the non-functional requirements to improve accuracy of software effort estimation through case based reasoning

Abstract

Producing accurate software effort estimation is essential for effective software project management that remains a considerable challenge to software engineering and software industry in general. Many methods have been proposed to increase the accuracy of estimating the software project size, effort, or cost. However, the primary focus has been on functional requirements FRs. We are convinced that the rigorous estimation requires a thorough knowledge of all the requirements of the software to be measured. Consequently, a clear identification of the FRs and NFRs as well as a strong understanding of the relationships existing between them is crucial to get measurements closer to reality. In this paper, we propose an early software size and effort estimation method based on a combination of COSMIC and case based reasoning that uses individual requirement measurements as a solution to improve the performance of CBR and to increase the precision of the estimations. This hybrid technique consists in adjusting the FRs measurements by the effect of NFRs with which they are connected. A new link requirements model is proposed in which the possible relationships existing between FRs and NFRs are expressed. This combination will help to efficiently include NFRs, and their relations with FRs, earlier in the measurement process and throughout the life cycle of the software development project.