The Use of Activity Models and Regression Analysis for Development Effort Forecasting
Nayan Ranjan Paul, Pulak Sahoo, Asit Kumar Das, Prateek Sahoo, Chandan Patra · 2024
Majority of modern-day industries operate with the help of computer applications which contain hardware and software parts. Software programs are extremely important and developing them can be quite challenging. Creation of high-calibre software, early prediction of work-effort is of paramount significance. This study attends to this requirement for modern-day applications. The process followed in this study extracts the features existing in the UML(Unified Modeling Language) Activity-models made for the systems. These features collaborated with custom-made Machine Learning Regression Analysis (ML-RA) programs were utilized for early prediction of work-effort. The ML-RA programs employed were: Bayesian-Ridge-Regression (BRR), Artificial-Neural-Network (ANN), K-Nearest-Neighbor (KNN), Extreme-Gradient-Boosting (EGB) and Extreme-Learning-Machines (ELM). Based on the outcomes of our experiments, it was clear that the ELM approach provided the highest accuracy with respect to other ML models.