Bayesian Network as a Decision Support System in the Company’s Risk Management System of Emergency Situations

Emergencies can inflict not only irreparable harm to individuals but also to a company’s operations. If we can foresee the onset of intricate emergency situations with catastrophic outcomes, we then have the chance to reduce these repercussions through the implementation of tailored risk management measures, safeguarding the company’s activities. Traditional methods of modeling and forecasting complex (extraordinary) natural processes do not yield dependable data for constructing viable scenarios in the analysis of strategies or making informed management decisions. Therefore, in the study, the authors proposed a scientific-methodical approach to the preventive orientation of risk management of emergencies of a company, which is improved by an oriented acyclic graph of the Bayesian network, which gives the posterior probability as a result and, unlike others, takes into account the causality of emergency events of the company. An algorithm for constructing a Bayesian network with a directed acyclic graph using the Netica software product is proposed. This tool - an acyclic graph of the Bayesian network - gives a posterior probability as a result and, unlike others, provides variability in the assessment of the impact of factors and considers the causality of a company’s extraordinary events. Approbation of the developed methodological and analytical toolkit proved its real nature. Modeling can be approached in both traditional “top-down” and “bottom-up” methods, depending on whether you start with pre-established model characteristics or build the model from the ground up. A deeper understanding of the occurrence and progression of emergencies, as well as the factors triggering their activation, can assist companies in devising management strategies that increase the likelihood of attaining and sustaining a healthy ecosystem. Conventional methods for modeling and forecasting complex (extraordinary) natural processes do not yield dependable data for creating viable scenarios when analyzing strategies and making informed management decisions. The article’s significance lies in its practical contribution to emergency risk management through Bayesian networks and its novelty in introducing a preventive approach tailored to specific emergency scenarios in companies. Additionally, it emphasizes the broader applicability of Bayesian networks.