Evaluation and prediction of time overruns in Jordanian construction projects using coral reefs optimization and deep learning methods
Within the complex domain of construction project management, the accurate anticipation of time overruns is a significant obstacle, particularly within the specific context of the construction sector in Jordan. This study aimed to utilize deep learning, specifically the Multi-Layer Perceptron (MLP),...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Published: |
Springer
2024
|
| Online Access: | http://psasir.upm.edu.my/id/eprint/105751/ https://link.springer.com/article/10.1007/s42107-023-00936-3?error=cookies_not_supported&code=b1ea9bc5-56ed-4791-b9cb-13359e60b686 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Within the complex domain of construction project management, the accurate anticipation of time overruns is a significant obstacle, particularly within the specific context of the construction sector in Jordan. This study aimed to utilize deep learning, specifically the Multi-Layer Perceptron (MLP), and enhance its overrun predictive ability by incorporating the Coral Reefs Optimization Algorithm (CROA). The approach employed in our study involved the utilization of a comprehensive dataset encompassing diverse aspects of building projects, ranging from financial indicators to project durations. The Multilayer Perceptron (MLP) was utilized as the underlying framework, and the model's parameters were refined utilizing the Covariance Matrix Adaptation Evolution Strategy (CMA-ES) with Rank-One Adaptation (CROA) to enhance the predictive capabilities. The results of our study indicate that the MLP-CROA continuously showed superior performance compared to the standalone MLP across many metrics. Notably, the MLP-CROA had particularly excellent values in metrics such as R-squared, which suggests its solid predictive capabilities. In conclusion, this study highlights the significant potential for transformation in the Jordanian construction industry by integrating deep learning and bio-inspired optimization methodologies. This process improves forecast accuracy and facilitates proactive project management, potentially directing projects toward timely and cost-effective completion. |
|---|