Project scheduling and performance prediction: A fuzzy-Bayesian network approach

Purpose: Despite the extensive research in project risk management and availability of several techniques and tools, quantifying uncertainty in project schedules remains a challenge. Current risk analysis models suffer from several shortcomings that need to be addressed to provide more reliable and valid schedules. This paper aims to present a dynamic decision support system with the purpose of providing project managers with necessary tool for making real-time informed decisions. Design/methodology/approach: The proposed approach incorporates the widely accepted critical path method (CPM) calculations in a Bayesian network (BN). BN is employed to conduct inferencing and causal analysis and provide probabilistic results, which can improve the decision-making process. Time parameters of each activity in the CPM network is modeled by a set of simulation nodes in the BN. Prior probability distribution of activities duration is extracted from experts using a fuzzy analytical solution. Findings: The model proposed in this paper is able to address some key outstanding issues of current project scheduling techniques through: (1) modeling the causality among different sources of schedule uncertainty, (2) minimizing uncertainty in experts' evaluations, (3) assessing effects of unknown risk factors and (4) using actual activity data for learning the behavior of project and predicting crew productivity. Originality/value: The purposed methodology provides a framework for the new generation of project schedule analysis tools that are better informed by available knowledge and data, and hence, more reliable and useful.