A New Approach in Software Cost Estimation Using Regression Based Classifier

The correct cost estimation for software companies and company executives and companies is very important task. In required effort prediction, estimated time of delivery and cost of the project with high accuracy is still vast challenge for projects masters. For many years, algorithmic models like Constructive Cost Model (COCOMO) families were used for required effort prediction. Nowadays, intelligent methods have many applications in SCE. In this study, we firstly give introduction to software cost estimation (SCE) and requirements. Then, we proposed a new intelligent approach by means of additive regression model for classifying the training and testing instances on NASA projects. The results show that our proposed system is more effective. Performance is evaluated by comparing the COCOMO and additive regression based classifier results. Keywords: Software Cost Estimation, Data Mining, COCOMO, Additive Regressions;