An Extreme Learning Machine based Approach for Software Effort Estimation

Software Effort Estimation (SEE) is the task of accurately estimating the amount of effort required to develop software. A significant amount of research has already been done in the area of SEE utilizing Machine Learning (ML) approaches to handle the inadequacies of conventional and parametric estimation strategies and align with present-day development and management strategies. However, mostly owing to uncertain outcomes and obscure model development techniques, only a few or none of the approaches can be practically used for deployment. This paper aims to improve the process of SEE with the help of ML. So, in this paper, we have proposed an Extreme Learning Machine (ELM) based approach for SEE to tackle the issues mentioned above. This has been accomplished by applying the International Software Benchmarking Standards Group (ISBSG) dataset, data pre-processing, and cross-validation. The proposed approach results are compared to other ML approaches (Multi-Layer Perceptron, Support Vector Machine, Decision Tree, and Random Forest). From the results, it has been observed that the proposed ELM based approach for SEE is generating smaller error values compared to other models. Further, we used the established approaches as a benchmark and compared the results of the proposed ELM-based approach with them. The results obtained through our analysis are inspiring and express probable enhancement in effort estimation.