Improved Software Cost Estimation using General Search based Analogous Project Feature Weight Assignment
Software cost estimation is the estimation of development effort and time required to develop any software project. Analogy based estimation is considered as one of the most popular techniques used in software cost estimation. Analogy based estimation is mostly appraised for its expensive computation cost , low prediction accuracy and large memory requirements despite many research works focusing on optimizing the weights of the project features being carried out in past for the overall purpose of improvement in performance of analogy based software cost estimation. In this paper, a general search based feature weight assignment has been proposed for improving the analogous based software cost estimation. Here in this proposed work, separate weights to each project feature and to find the optimal weights by general search is done. The proposed approach has been exercised on several real-world data sets like Albrecht and Desharnais and evaluated using commonly used quality metrics like Mean of Magnitude of Relative Error (MMRE). The promising results achieved from the study were found to increase estimation accuracy and reliability which likely increases the models chance of acceptance too.
- There are currently no refbacks.