<![CDATA[Software effort estimation is one of the critical aspects of software project management, but it often faces accuracy issues. Although statistical methods such as Linear Regression have been used, previous research has shown that these models are often inefficient because they involve many variables that may not be relevant. This study aims to improve the performance of Linear Regression models in software effort estimation using Forward Selection feature selection techniques. Two models were compared: the conventional Linear Regression model and the model with Forward Selection. Evaluation metrics include Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Square Error (RMSE), and coefficient of determination (R2). Results show significant improvements in all performance metrics on models with Forward Selection. Notably, the MSE increased from 1.0 to 0, suggesting that this model is more effective in explaining data variability. The use of Forward Selection in Linear Regression models for software effort estimation shows significant performance improvements and is worthy of consideration for further industry research and practice.]]>
Copyrights © 2024
