Automated Prediction of Construction Project Efficiency Rating Using Multi-Layered Neural Network with Sigmoid Function

  • Mary Mae S. Pontillas, Dante L. Silva, Kevin Lawrence M. De Jesus

Abstract

 Time overrun is defined as excess or extra time required to complete the agreed scope of works following what is intended during project planning, or at the beginning of the project. Time overrun is relative in cost overrun, therefore reducing either or both of them is the main driver of any project and construction managers. The research aims to conduct a study regarding the utilization of automated prediction of construction project efficiency to understand the relationship amongst change order, time overrun and project efficiency and predict potential mitigating measures and used as decision-making supplement once factors were observed during project execution. The data collected were analyzed using a multi-layered neural network with sigmoid function using internal parameters of Levenberg-Margquadt for the Training Algorithm, hyperbolic tangent sigmoid (tansig) transfer function and 12 hidden neurons. The questionnaire tool was analyzed on its reliability using the coefficient of internal consistency (Cronbach α) and resulted in an excellent consistency rating. Results indicated that among factors of a change order, the client has the most frequency that affects change order followed by the consultant and finally by the contractor. Also, time overrun was mostly affected by excusable & non-compensable factors which imply of factors that are unforeseeable by all parties, however not compensable as provided in contract clauses. Then followed by the factors of non-excusable & non-compensable and finally by excusable & compensable. Accordingly, between change order and time overrun, project efficiency was mostly affected by time overrun factors. The study generated an automated prediction model using a multi-layered neural network and resulted in an MSE and MAPE of 0.042542 and 6.80846% respectively, which indicates a good model in predicting a construction project efficiency rating. The study concluded by recommending that the tool made in this study be provided with an automated to help the beneficiary in the construction field.

Published
2020-06-01
How to Cite
Mary Mae S. Pontillas, Dante L. Silva, Kevin Lawrence M. De Jesus. (2020). Automated Prediction of Construction Project Efficiency Rating Using Multi-Layered Neural Network with Sigmoid Function. International Journal of Advanced Science and Technology, 29(08), 2631 - 2639. Retrieved from http://sersc.org/journals/index.php/IJAST/article/view/23441
Section
Articles