Integrating Non-Functional Requirements Into a Case Based Reasoning Model for Software Effort Estimation

Abstract

The importance of accurate effort estimates is increasing in the software development industry. As a result, many methods have been proposed in order to deal with the inaccuracy and imprecision of software effort estimation. However, the most of them focus on functional requirements (FR), while non-functional requirements (NFRs) are often ignored or reflect only a very high level of features required, causing the estimates to be increasingly inaccurate. Therefore, there is a need to better understand and include all the characteristics of the software to be measured. In this paper, we are going to apply machine learning using case based reasoning (CBR) model, combined with COSMIC, to improve the precision of the estimates. This hybrid technique uses COSMIC to measure the functional size of FRs, takes into account the relationships between FRs and NFRs when measuring the impact of NFRs on the effort of FRs with which they are associated. The relationships existing between FRs and NFRs are identified in the link requirements model proposed. We aim, by this combination, to reduce the uncertainty of estimates by including all types of requirements in the measurement process.