AI

Accuracy of Contemporary Parametric Software Estimation Models: A Comparative Analysis

Abstract

Predicting the effort, duration and cost required to develop and maintain a software system is crucial in IT project management. Although an accurate estimation is invaluable for the success of an IT development project, it often proves difficult to attain. This paper presents an empirical evaluation of four parametric software estimation models, namely COCOMO II, SEER-SEM, SLIM, and TruePlanning, in terms of their project effort and duration prediction accuracy. Using real project data from 51 software development projects, we evaluated the capabilities of the models by comparing the predictions with the actual effort and duration values. The study showed that the estimation capabilities of the models investigated are on a par in accuracy, while there is still significant room for improvement in order to better address the prediction challenges faced in practice.