Quantifying Risks using Bayesian Networks

M. Fineman and N. Fenton (UK)

Keywords

Risk theory and uncertainty, Risk assessment approaches and methodologies, Risk management, Risk events, Opportunity events, Bayesian Networks

Abstract

This paper concentrates on the perspective of a decision maker at the start of the decision making process. At the start of every project managers discuss and list risks. Many people, who consider themselves to be at the state of-the-art in quantified risk assessment, then define risk as the measure: Risk = Probability x Impact. This approach has fundamental problems. Not only it is difficult to elicit ‘probability’ and ‘impact’ in many cases it’s meaningless, and it also leads to a paradox. When building a risk register the more we think about risks apparently the greater the risk of the project. The solution we propose is based on Bayesian Networks (BNs), which capture a proper causal view of risk events. The sensible risk measures that we are proposing are simply the probabilities you get from running a Bayesian Network. The new framework has a dual approach for quantifying opportunities and their consequences. Risk events and opportunity events mirror each other. Hence everything we said about risk events applies to opportunity events. Therefore, an immediate impact of this work is to improve and unify risk quantification by incorporating opportunities in modelling.

Important Links:



Go Back