Optimizing software cost estimation with a novel adaptation-based approach using evolutionary algorithms

In mathematical models and applications, evolutionary algorithms have proven indispensable for solving optimization problems. These algorithms may, however, eventually experience many issues, including diversity loss and stagnation. To overcome these challenges, a novel mutation operator has been introduced and incorporated into the differential evolution (DE) algorithm. This newly proposed mutation operator, based on the homeostasis factor, not only ensures global search and environmental balance but also promotes diversity within the problem.The effectiveness of the proposed method has been validated through two tests: experimentation and comparative studies on benchmark functions, as well as cost estimation for software applications. In the experimentation and comparative study, the proposed method achieved satisfactory results, with the minimum number of function evaluations and statistical functions reaching the target value of 108. Additionally, the proposed operator demonstrated favorable outcomes in terms of minimum error for each function, as confirmed by the Wilcoxon rank-sum test (p = 0.05). The experimental results indicate that the proposed method outperforms state-of-the-art algorithms, offering superior outcomes in terms of minimum error, accuracy prediction, mean magnitude of relative error (MMRE), and root mean square error (RMSE).