Predicting the Impact of Project Bundling Objectives under the Design Build (DB) Project Delivery Method Using Supervised Machine Learning

Current infrastructure systems are aging and require immediate actions. Project bundling has introduced as an innovative project delivery approach that groups several infrastructure projects into a single contract. However, research on project bundling is still considered in its early stages. This research develops guidelines for implementing project bundling using the Design Build (DB) delivery method through predicting the impact of project bundling objectives on the overall bundle program using machine learning algorithms. First, a survey was distributed to collect expert opinions. Second, survey results were preprocessed for further analysis. Third, three classification algorithms were implemented, including K-nearest neighbors, random forest, and artificial neural networks (ANN) algorithms. The results showed that the ANN model has the highest prediction accuracy of 90.91%. Ultimately, this research helps decision-makers prioritize the objectives that have the highest influence on their bundling program and thus optimize their bundling practices under the DB method.