A Model-Driven Decision Support System for Software Cost Estimation (Case Study: Projects in NASA60 Dataset)

Abstract

Estimating the costs of software development is one of the most important activities in software project management. Inaccuracies in such estimates may cause irreparable loss. A low estimate of the cost of projects will result in failure on delivery on time and indicates the inefficiency of the software development team. On the other hand, high estimates of resources and costs for a project will waste opportunities for other projects. This paper presents a methodology for estimating the costs of software development. The methodology is a model-driven decision support system that consists of four subsystems; namely Data subsystem, Model subsystem, User Interface subsystem, and Knowledge subsystem. The core supports of this system are based on coherent theory on the nature of collaborative work and their mathematical models in software engineering that included in the model subsystem. This core provides a theoretical foundation for decision optimizations on the optimal labor allocation, the shortest duration determination, and the lowest workload effort and costs estimation. The experimental results and evaluations on Dataset NASA60 show that the proposed system has significant conformance with experience in practice. Based on the proposed decision support system, a wide range of fundamental problems in software project organization and cost estimation can be solved rigorously.