علي رمضاني

Timely project delivery and cost control within the set budget are considered to be the most important indicators of the success of any project. Failure to achieve any of these goals can lead to the failure of the project and failure to achieve the final goals. Therefore, the most important task of the project manager is to manage time and cost in the project implementation phase in such a way as to ensure that the project is completed on time and with the budget. This important thing requires a tool that can be used to estimate the time and cost of completing the project. Among the project progress forecasting methods, the Estimating at Completion (EAC) tool, which is one of the Earned Value Management (EVM) tools, is very common and widely used. Project managers use EAC to estimate the final project time and cost during project execution. When it is seen that the time and cost of the project will be much more than what was initially considered and the so-called progress, you will deviate too much from the plan. They consider the required corrective measures and apply them at the right time and part of the project. So that the project returns to its original path and is completed according to the planned cost and time. However, the EVM method faces limitations that can reduce the accuracy of predictions. For example, traditional EVM models often assume that costs change linearly, which in many cases is incorrect. Also, this method predicts the future performance of the project based on the past results, if there is no guarantee that the future results will be in perfect agreement with the past performance in the environment full of uncertainty of the project. Considering the poor performance and accuracy of earned value management and proving the effectiveness of machine learning methods in forecasting and considering that artificial intelligence (AI) techniques and data analysis can improve managerial decision-making in conditions of uncertainty. Therefore, in this research, using machine learning methods (Lasso, Ridge, SVM, decision tree and random forest) and based on the collected data of a set of projects, models have been developed to predict the cost and time of the project. The performance of the mentioned models has been measured with the help of MAPE, RMSE and R2 eva‎luation criteria, and the results of the models have been compared with the results of the EAC method, and the results indicate the good performance of the proposed models compared to the EAC method. Among the proposed models, the best accuracy in predicting time is related to the Lasso model with RMSE=12, MAPE=4% and R2=.97. In addition, the decision tree model with RMSE=145270, MAPE=5% and R2=.95 has the best accuracy in predicting the cost of projects.

پروژه , مديريت ارزش كسب شده , برآورد درتكميل , يادگيري ماشين , جنگل تصادفي , ماشين بردار پشتيبان

Project , Earned value management , Estimate at completion , Machine learning , Random forest , Support vector machine

ali ramezani

Dr siamak noori