A Novel AI-Enabled Metaheuristic Optimization Framework for Accurate Software Cost Estimation Using Benchmark and Industrial Datasets

Abstract

Accurate software cost estimation is a significant aspect in software project management because it has a direct impact on software projects. However, software cost estimation techniques have some limitations in managing high-dimensional software project variables and nonlinear relationships between software cost drivers. These limitations often lead to inefficient software project management and inaccurate software cost estimation. In order to overcome these limitations and inaccuracies in software cost estimation techniques, this paper proposes a novel AI-based metaheuristic optimization technique for software cost estimation by integrating more accurate machine learning models and intelligent optimization techniques.The proposed software cost estimation model incorporates a Transformer regression model and a Genetic Algorithm (GA) to efficiently estimate software cost. The significant software project characteristics used in this proposed software cost estimation model are project size in KLOC, function points, requirement stability, team size, developer experience, productivity index, system complexity, technology maturity, and tool automation level. Performance of model is measured by common evaluation metrics such as Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Magnitude of Relative Error (MMRE). The experimental outcome shows that proposed hybrid model has significant effect improving the accuracy of software cost prediction compared to the existing machine learning and estimation techniques. The model shows an accuracy of about 92% with less estimation errors, proving its efficiency in modeling complex project characteristics. The proposed AI-powered metaheuristic optimization model presents a robust and efficient solution for intelligent software cost estimation.