This paper introduces the integration of the Social Group Optimization(SGO)algorithm to enhance the accuracy of software cost estimation using the Constructive Cost Model(COCOMO).COCOMO’s fixed coefficients often lim...This paper introduces the integration of the Social Group Optimization(SGO)algorithm to enhance the accuracy of software cost estimation using the Constructive Cost Model(COCOMO).COCOMO’s fixed coefficients often limit its adaptability,as they don’t account for variations across organizations.By fine-tuning these parameters with SGO,we aim to improve estimation accuracy.We train and validate our SGO-enhanced model using historical project data,evaluating its performance with metrics like the mean magnitude of relative error(MMRE)and Manhattan distance(MD).Experimental results show that SGO optimization significantly improves the predictive accuracy of software cost models,offering valuable insights for project managers and practitioners in the field.However,the approach’s effectiveness may vary depending on the quality and quantity of available historical data,and its scalability across diverse project types and sizes remains a key consideration for future research.展开更多
文摘This paper introduces the integration of the Social Group Optimization(SGO)algorithm to enhance the accuracy of software cost estimation using the Constructive Cost Model(COCOMO).COCOMO’s fixed coefficients often limit its adaptability,as they don’t account for variations across organizations.By fine-tuning these parameters with SGO,we aim to improve estimation accuracy.We train and validate our SGO-enhanced model using historical project data,evaluating its performance with metrics like the mean magnitude of relative error(MMRE)and Manhattan distance(MD).Experimental results show that SGO optimization significantly improves the predictive accuracy of software cost models,offering valuable insights for project managers and practitioners in the field.However,the approach’s effectiveness may vary depending on the quality and quantity of available historical data,and its scalability across diverse project types and sizes remains a key consideration for future research.
