Forecasting Company's Bankruptcy Using Neural Networks

This paper is devoted to forecasting companies’ bankruptcy using artificial neural networks. Despite the fact that the problem of prediction is standard for neural networks, which can be successfully solved by them, neural networks are not widespread in question of forecasting in comparison with classical econometric models of prediction: regressions. At the same time, artificial neural networks are able to show more qualitative results comparing with regressions. This fact is also reflected in this paper. This research examined financial information for 100 companies. A half of them became bankrupt one year later. Using Rstudio interface a neural network was successfully modeled, which showed clear and high result of bankruptcy forecasting. Obtained results were compared with results of bankruptcy forecasting using linear regression. This comparison reveals the fact that results showed by artificial neural network are better than results of bankruptcy forecasting made by linear regression.