Application of Feed Forward Multilayer Neural Network and Recurrent Neural Network in the Prediction of Business Insolvency – Comparative Analysis
This article is about prediction of the bankruptcy of enterprises in Central Europe. The objective of the conducted research is to develop a predictive model using the feedforward multilayer neural network and recurrent neural network for the process of forecasting the risk of corporate bankruptcy. The effectiveness of the programmed artificial neural network models was verified with the use of two separate forecasting horizons – one year and two years before the bankruptcy of enterprises. The author carried out two research approaches characterized by different selection of input variables - minimization (with the use of only four financial ratios) and maximization (with the use of 28 different variables). The article presents also comparative analysis of the results of artificial neural networks from these two research approaches.
Index Terms - Artificial intelligence, bankruptcy, forecasting. The study has been prepared within the grant project No. 2015/19/B/HS4/00377, "Trajectories of life and the collapse of companies in Poland and in the world - identification, evaluation and forecast." Research funded by the National Science Centre in Poland (Narodowe Centrum Nauki)