A Comparison of Development Effort Estimation Techniques for Web Hypermedia Applications

Several studies have compared the prediction accuracy of different types of techniques with emphasis placed on linear and stepwise regressions, and Case-based Reasoning (CBR). We believe the use of only one type of CBR technique may bias the results, as there are others that may also be used for effort prediction. This paper has two objectives. The first is to compare the prediction accuracy of three CBR techniques to estimate the effort to develop Web hypermedia applications. The second objective therefore, is to compare the prediction accuracy of the best CBR technique, according to our findings, against three commonly used prediction models, namely multiple linear regression, stepwise regression and regression trees. One dataset was used in the estimation process and the results showed that different measures of prediction accuracy gave different results. MMRE and MdMRE showed better prediction accuracy for Multiple regression models whereas boxplots showed better accuracy for CBR.