Specification of a Hybrid Effort Estimation System using UML

Abstract

Software effort estimation (SEE) predicts the amount of effort needed to develop software and this information is important for a project manager to make good decisions when confronted with requirement change requests during the development. However, SEE can predict accurate development effort only if the project data is complete and stable. In real project, this is not the case and becomes a very challenging issue. Inaccurate effort estimation will result in lots of difficulties in managing a software project, hence the project’s failure rate increases. Both algorithmic and nonalgorithmic methods in SEE have their respective advantages and limitation. This paper presents a hybrid approach combining algorithmic and non-algorithmic SEE methods to propose a more accurate decision to the project manager when a requirement change is requested by a stakeholder. This hybrid model is translated into an effort estimation system that allows software estimation experts to intervene in the numerical estimation calculation to propose a more accurate effort estimation to help the project manager make a better decision to accept or not the change request. In this paper Unified Modeling Language (UML) is used to prepare the software specifications of the hybrid system, resulting in the complete documentation of the system’s Software Requirement Specifications (SRS).