Accurately estimating software costs is a vital step in ensuring the successful completion of a software project. There is a need for estimation techniques that ensure projects are completed on time, within budget, and with the desired quality. Accurate estimation plays a crucial role in crafting realistic budget plans and ensuring that projects are completed on time with sufficient resources. When estimations are precise, teams can spot potential issues early, distribute resources more effectively, and handle risks with greater confidence. This research focuses on boosting the reliability of software effort estimation by applying a stacking method enhanced with Bayesian hyperparameter optimization. It leverages three core algorithms: SVM, Random Forest, and Decision Tree, each fine-tuned using the proposed approach. Evaluations across 11 public datasets reveal noteworthy improvements, ranging from 0.2 to 0.5. A significance test confirms the model’s strong performance, showing a p-value greater than 0.5, which indicates that the results are statistically meaningful. These findings suggest that combining stacking with Bayesian tuning holds promise for refining software effort predictions. It can serve as a valuable reference for future project planning across diverse modeling approaches.
Yusuf et al. (Thu,) studied this question.