Estimation of project completion time using proper fuzzy combination of regression methods

One of the important issues in project management is estimation of project completion time. Moreover, dynamic process of project progress and the use of new knowledge in this procedure are vital issues in estimation of project completion. By considering this dynamic process and Earn Value Management (EVM) features in project management, a decision base from estimation models has been presented in this paper. This method is presented by clustering samples in the fuzzy way and by providing an ensemble model from estimators which use the possession rate in each cluster. Therefore, final estimation rate not only uses the combination of all estimators´ results, but also imports a degree of fuzziness in decision making which plays a significant role in uncertainty. One of the obvious features of this method is higher reliability that this method makes in comparison with every individual existing method in the ensemble. The second important feature of this method is its robustness against the existence of a weak estimator in ensemble. Therefore, the proposed method is in a way that the existing weak estimators in the ensemble have little impact on final results. Furthermore, controlling the number and the kind of existing regression models in ensemble is another main property of the proposed model. By employing stronger models in this ensemble, the rate of accuracy and reliability can be remarkably increased.