Bayesian Multi-Scale Optimization For Software Cost Estimation

Software Cost Estimation is very important challenging task for completing the project successfully. The estimation in software development depends on various factors particularly managing project cost, time and quality and effort factors. Therefore, accurate assessment is a consequential factor in projects success and reducing the risks. In the last two decades, many researchers and practitioners presented statistical and machine learning-based models for software effort estimation. In this paper, a novel approach based Bayesian multi-class algorithm is proposed for software cost estimation. It helps project manager to provide nimble and realistic estimate for the project effort and development time that in turn gives software cost. The proposed work is carried in two steps, in first phase known as training phase, optimizing the parameters and second step known as validating phase, the prediction process. The Parameters SSE, RME, MAD and R2are calculated for COCOMO-II dataset. Statistical results show that our method could significantly improve accuracy, error minimization and has potential to become an effective method for software cost estimation.