Fuzzy Inference System and Fuzzy Neural Inference System Applied to Risk Matrix Classification in Projects

Abstract

Projects are essential for organizations to transform strategies into results, but uncertain events can impose risks to achieve a certain objective. Risk management aims to support an organization in deciding how to deal with risks, prioritizing them through the application of Risk Matrices (RMs). RMs or Probability and Impact Matrices is used to support decision-making, helping management to classify and prioritize risks to decide which will be ad-dressed, monitored, or tolerated. RMs are supposedly easy to build and explain, but according to the literature they may contain uncertainties. To deal with uncertainty, it is recommended to apply a Fuzzy Inference System, based on Fuzzy Set Theory (FST) or a Fuzzy Neural Inference System with the presence of an artificial neural network. Thus, the aim of this paper was to develop and apply a Fuzzy Inference System (FIS) and a Fuzzy Neural Inference System (FNIS) in the classification of MRs in projects to reduce uncertainty. The analysis of the results indicated that the application of the two systems resulted in a continuous classification rule by smoothing the boundary areas between each of the RM classes, reducing uncertainty and improving risk classification. Both systems showed good results in reducing uncertainty. However, the results obtained with FNIS were more consistent. The main contribution of this work lies in the possibility of improving the decision making by reducing the uncertainty present in RMs.