A min-max Approach for Improving the Accuracy of Effort Estimation of COCOMO

Abstract

This study proposed to extend the Constructive CostModel (COCOMO) by incorporating the concept of min-maxapproach to estimation. Formal effort estimation modelslike Constructive Cost Model (COCOMO) are limited bytheir inability to manage uncertainties and imprecisionsurrounding software projects early in the developmentlife cycle. A min-max approach is suggested to rectifydata uncertainties and modeling errors. The proposedmethod of min-max is used to improve the accuracy ofeffort estimation of COCOMO and its result have beencompared with the gradient descent, robust fuzzyclustering, k-mean clustering methods of estimation. Ithas been observed that the proposed method have lowestand steady state absolute estimate error AE(k) and meanabsolute estimate error MAE(k) for different value ofk(time series) and different step-sizes.