Predicting the failure of institutions based on qualitative parameters using Bayesian Networks

Abstract

In the year 2020, the situation took dramatic contours for companies and entrepreneurs. With the COVID-19 pandemic, bankruptcy filings increased by about 12.7% compared to the previous year, and in addition, it was the first increase since the 2016 crisis. In this scenario, it is necessary to predict the bankruptcy of companies to anticipate actions and minimize the effects. The present work uses a tool known as Bayesian Networks, which is based on graph theory and probability theory to model an uncertainty scenario. The Network for the proposed study was manually modeled through the use of Netica, consequently, it was possible to obtain satisfactory results through the analysis of some sectors within the institutions. With the help of the tool, it was possible to predict the bankruptcy based on qualitative parameters of specialists, analyzing the risks that can result in the failure of organizations.