Empirical Software Engineering in Practice (IWESEP), 2014 6th International Workshop on

BACKGROUND: Several studies in software effortestimation have found that it can be effective to use a window ofrecent projects as training data for building an effort estimationmodel. The previous studies evaluated the use of a windowwith popular estimation models: linear regression (LR) andestimation by analogy (EbA). Many effort estimation models havebeen proposed, and the generality of windowing approach stillremains uncertain for other effort estimation models, especiallyfor those based on different theory. OBJECTIVE: This studyinvestigates the effect of using a window on estimation accuracywith Classification and regression trees (CART). CART wasrecently found as a good performance method, and is based ona different theory from LR and EbA. METHOD: We comparedthe estimation accuracy of a windowing approach and growingapproach with the same data set and procedure as the past stud-ies. RESULTS: There is a difference in the estimation accuracybetween using a window and not using a window. However, theeffctive range of using windows on CART is narrower than thaton LR. CONCLUSIONS: Windowing is also effective with CART.However, the range of effectiveness is narrower. The resultscontribute to the generality of the effectiveness of windowingapproach