Development and evaluation of a framework for software effort estimation

Abstract

The prediction of reliable estimates is very crucial for software development success. Having realistic estimates at an early stage a project’s life cycle project managers to provide the project’s budget, cost and to perfectly allocate financial and human resources. Despite the numerous effort estimation techniques and approaches that have been proposed in the literature, the estimates have remained inaccurate. In this dissertation, we aim to improve the accuracy of effort estimates; we explore new ways to reduce the uncertainty of estimates based on CBR (Case Based Reasoning). To cope with this challenge, we present a novel framework for software effort estimation, which uses a case-based reasoning (CBR) model, and considers a comprehensive set of different types of project parameters and requirements-functional requirements (FRs), non-functional requirements (NFRs), constraints and domain properties (DPs), while taking into consideration relationships among them. We believe that being able to understand requirements to be measured correctly is the basis for effort estimation activities. Moreover concerning the use of CBR, this framework offers a multiple-criteria decision analysis technique (MCDA) for enhancing the accuracy of similarity measures among cases of multiple past projects that are similar to the current software project, towards determining and selecting the most for effort estimation activities. Moreover concerning the use of CBR, this framework offers a multiple-criteria decision analysis technique (MCDA) for enhancing the accuracy of similarity measures among cases of multiple past projects that are similar to the current software project, towards determining and selecting the most similar one among them. What is more, our approach employs an incremental methodology of estimation in the different requirements gathering and development cycles, hence providing not a one-shot final estimate but instead to each cycle, as in the early stages of project development process requirements are not yet available, incomplete or with high degree of uncertainty.