IPTEK The Journal of Technology and Science, 34(2), 2023 (e/pISSN:2088-2033, 0853-4098) DOI: 10.12962/j20882033.v34i2.16693 Received 5 May, 2023; Revised 23 May, 2023; Accepted 31 May, 2023 ORIGINAL RESEARCH PROBABILISTIC SCHEDULING BASED ON HYBRID BAYESIAN NETWORK–PROGRAM EVALUATION REVIEW TECHNIQUE Tri Joko Wahyu Adi* | Farida Rachmawati | Safra Yulia Rizky Dept. of Civil Engineering, Institut Teknologi Sepuluh Nopember, Surabaya, Indonesia Correspondence *Tri Joko Wahyu Adi, Dept of Civil Engineering, Institut Teknologi Sepuluh Nopember, Surabaya, Indonesia. Email: trijokowahyuadi@gmail.com Present Address Gedung Teknik Sipil, Jl. Taman Alumni, Surabaya 60111, Indonesia Abstract Project scheduling based on probabilistic methods commonly uses the Program Evaluation Review Technique (PERT). However, practitioners do not widely utilize PERT-based scheduling due to the difculty in obtaining historical data for similar projects. PERT has several drawbacks, such as the inability to update activity dura- tions in real time. In reality, changes in project conditions related to resources have a highly dynamic nature. The availability of materials, fuctuating labor productiv- ity, and equipment signifcantly determine the project completion time. This research aims to propose a probabilistic scheduling model based on the Hybrid Bayesian Network-PERT. This model combines PERT with Bayesian Network (BN). BN is used to accommodate real-time changes in resource conditions. The modeling of BN diagrams and variables is obtained through an in-depth literature review, direct feld observations, and distributing questionnaires to experts in project scheduling. The model is validated by applying the proposed model to a 60 m concrete bridge construction project in Indonesia. The simulation results of the proposed model are then compared with the case study project to assess the model’s accuracy. The result of the study shows that the proposed hybrid Bayesian-PERT model is accurate and can eliminate the weaknesses of the PERT method. Besides being able to provide an accurate prediction of project completion time (93.4%), this model can also be updated in real-time according to the actual condition of the project. KEYWORDS: Bayesian Network, Construction Risk, PERT, Productivity, Project Scheduling