Decision Support for Project Management by using Diagnostic Inference and Explaining-Away in Bayesian Networks
Barboros Yet, Erhan Pişirir · OpenMETU (Middle East Technical University) · 2020
Risk and uncertainty are natural elements of projects. A project manager has to manage these elements effectively to improve the chance of project success. Yet, well-known biases and heuristics about decision-making under uncertainty limit a project manager's ability to deal with these elements, especially for complex projects, and decision support tools can be helpful for this task. This paper focuses on building a decision support model for reasoning with uncertainty in project control. We propose a Bayesian Network (BN) model that uses the notions of diagnostic reasoning and 'explaining-away' to infer the effort requirements, progress and risks of project tasks. The proposed model aids the project manager in reasoning with uncertainty and risk factors when monitoring project progress. We use a project example to demonstrate the model structure, its use, benefits and limitation compared to conventional project control tools.