Using Bayesian Modeling to Forecast Resource Demand in the Planning Phase of Construction Projects

When embarking on a construction project, it is important to forecast when and how much resource demand will be required in the planning phase to ensure success. This study first examines expenditures on construction projects as a proxy for resource demand. The examination reveals that methods for forecasting resource demand are required to consider data variability. The examination shows that it is valuable to identify representative patterns of resource allocation along project timelines. However, conventional methods rarely allow simultaneous consideration of variability and allocation patterns. To address this limitation, this study presents a Bayesian statistical method that integrates a multiplier approach with the S-curve method. This method is evaluated by comparing the forecasts with real-world data. The results demonstrate that this method allows the simultaneous consideration of variability and allocation patterns. Furthermore, the method contributes to improvements in the explainability of resource demand forecasts because the model enables practitioners to express forecasts using the concept of a resource multiplier, which is often used in practical and conventional procedures.