Software Effort Estimation through a Generalized Regression Neural Network

Management of large software projects includes estimating software development effort as the software industry is unable to provide a proper estimate of effort, time and development cost. Though many estimation models exist for effort prediction, a novel model is required to obtain highly accurate estimations. This paper proposes a Generalized Regression Neural Network to utilize improved software effort estimation for COCOMO dataset. In this paper, the Mean Magnitude Relative Error (MMRE) and Median Magnitude Relative Error (MdMRE) are used as the evaluation criteria. The proposed Generalized Regression Neural Network is compared with various techniques such as M5, Linear regression, SMO Polykernel and RBF kernel.