Towards a correction factors-based software productivity using ensemble approach for early software development effort estimation

Abstract

Accuracy of effort estimation is one of the necessary conditions for efficiently managing software development projects. Since the information available in the early stages of software development is insufficient, software sizing metrics are considered critical factors for effort estimation. However, there is no consistent method for converting software sizing into the corresponding effort. Previous estimation methods have not considered software productivity a critical factor in estimating effort based on software sizing. This paper proposes a software productivity model based on correction factors in the Optimizing Correction Factors method through an ensemble construction mechanism of three popular machine learning techniques. The results show that using the proposed software productivity minimizes the estimation error of the methods compared to using fixed productivity metrics.