Particle Swarm Optimization-based Linear Regression for Software Effort Estimation

Abstract

In the context of software effort estimation, this study investigates the use of Particle Swarm Optimization (PSO)-based Linear Regression to improve estimation accuracy. The main problem faced is the limitations of standard Linear Regression models in accurately estimating the effort required for software development projects. This research aims to improve the quality of estimation of software efforts to optimize resource management and project schedules. The method used was the integration of PSOs in Linear Regression, which was evaluated using three different COCOMO datasets. Experimental results show that LR+PSO models consistently outperform standard Linear Regression with lower MAE, MSE, and RMSE, as well as higher R-squared. In conclusion, integrating PSOs in Linear Regression effectively improves the estimation accuracy of software efforts, demonstrating great potential for improving estimation quality in software project management practices.