Ensembling Teaching-Learning-Based Optimization Algorithm with Analogy-Based Estimation to Predict Software Development Effort

Accurate evaluation of software development effort estimation is a significant challenge for project planning and management. The analogy-based estimation (ABE) model is an effective method to estimate the effort required for software development. In this study, we proposed a teaching-learning-based optimization (TLBO) guided analogy-based estimation (TABE) model in which TLBO is ensembled with ABE to estimate effort in a reliable way. TLBO generates optimized feature weights during training, and those optimized feature weights are used during testing to estimate development effort. Five datasets, including Kemerer, Desharnais, Albrecht, China and Maxwell, were used to validate the proposed TABE algorithm. The experimental findings demonstrate that TABE can provide a better PRED while minimizing the MMRE value for a few of the studied datasets.