A Deep Learning Approach for Forecasting Cost Estimate at Completion (EAC) in Construction Projects

• Author / Creator

  Denisse Magdalena Diaz Merino

• Inaccurate cost forecasting is a significant issue that can lead to potential budget overruns, cash flow problems, poor stakeholder relationships, and financial losses for construction execution companies. To improve cost forecasting accuracy, this research proposes a deep-learning framework structured into five phases, starting with exploring the literature review and finishing with the practical application of the developed model. This framework leverages historical data from completed projects and deep-learning algorithms to enhance the cost estimate at completion accuracy.
  First, this study explores the literature on previous approaches for cost forecasting and examines factors that influence cost forecasting. Second, it analyses current practices in the industry, gathers historical data, and establishes a data acquisition model at the level of individual work packages. Third, it develops a computerized model, including data preprocessing, designing, and building forecasting models based on deep learning algorithms. It also develops a graphical user interface (GUI) to store generated data and deploy the deep learning model. Fourth, model application and verification are performed using the dataset to select the optimal forecasting algorithm and compare results with traditional earned value methods. Finally, the GUI is applied to deploy the cost forecasting model at the work package level.
  The study results demonstrate that the Gated Recurrent Unit (GRU) algorithm significantly outperforms traditional cost forecasting methods. The GRU model achieved Mean Absolute Percentage Errors (MAPE) of 7.38%, 6.14%, and 3.97% for the concrete, backfill, and piping work packages, respectively. In contrast, the Earned Value Management (EVM) method yielded MAPE values of 11.32%, 13.42%, and 13.3% for the same work packages. This study demonstrates the effectiveness of the deep learning model in accurately predicting cost forecasting for ongoing construction projects. By integrating deep learning algorithms with a comprehensive analysis of historical cost data, the proposed framework offers a methodological foundation for future innovation in cost forecasting analytics.

• Subjects / Keywords

  • Machine Learning
  • Deep Learning
  • Time series
  • Forecasting
  • Estimate at Completion (EAC)
  • Project Cost Management
  • Construction Projects
  • work package

• Graduation date

  Fall 2024

• Type of Item

  Thesis

• Degree

  Master of Science

• DOI

  https://doi.org/10.7939/r3-gny6-qd60

• License

  This thesis is made available by the University of Alberta Library with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.