Application of artificial neural networks for prediction of business indicators

  • 1 University of Chemical Technology and Metallurgy – Sofia Bulgaria


This paper examines the applicability of the neural networks in developing predictive models. A predictive model based on artificial neural networks has been proposed and training has been simulated by applying the Long Short-Term Memory Neural Network module and the time series method. Python programming language to simulate the neural network was used. The model uses the stochastic gradient descent and optimizes the mean square error. Business indicators for forecasting the results of the activity and the risk of bankruptcy of a company are forecasted and a comparison of the obtained forecast values with the actual ones is performed in order to assess the accuracy of the forecast of the developed model. As a result, it can be noted that business indicators can be successfully predicted through the Long Short-Term Memory Neural Network and the forecasted values are close to the actual ones.



  1. Altman E., Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy, Journal of Finance,09/1968, pp. 589-609.
  2. Simeonov O., N. Petrova, Controlling, basics of controlling, operational controlling, strategic controlling, Publishing Complex - UNWE, Sofia, 2012, pp. 191.
  3. Kasarova V., The New Metrics of Corporate Financial Success, Ed. Evdemonia Production EOOD, Sofia, pp. 24.
  4. Hristov I., Integrated model for evaluation of the activity of the enterprise, “Isvestia” - Publication of the University of Economics - Varna, pp. 66-77
  5. Kasarova V., Models and indicators for analys is of the financial stability of the company. Scientific electronic archive of NBU.
  6. Popchev I., Lectures in Computer Science and Technology - IICT-BAS, №1, 2016, Six topics in risk management,
  7. Deneva I., Annual Activity Report - Intersection of Science and Business in the Financial Statements, Department of Accounting and Analysis, Faculty of Finance and Accounting at UNWE, pp. 8-9
  8. Chabaan A., Y. Malevergne, J. Laurent, F. Turpin, Alternative risk measures for alternative investments, AFFI 2003, pp. 9- 21
  9. Kabaivanov St., Improving the accuracy of financial analysis with the help of neural networks, Reports from a scientific conference with international participation, Advanced scientific tools in management, 19. 09. 2008 - 23. 09. 2008, Ravda, Volume 1/2010 ISSN 1314-0582, pp. 247-253
  10. Laptev N., J. Yosinski, Li Erran, Li Slawek Smyl, Time-series Extreme Event Forecasting with Neural Networks at Uber, ICML 2017, Time Series Workshop, Sydney, Australia
  11. Borovykh A., S. Bohte, W.Cornelis, W. Oosterlee, Conditional Time Series Forecasting with Convolutional Neural Networks, Cornell University
  12. Chatfield C., The analysis of time series: an introduction, 2016
  13. Tsvetanova E., Adaptation of the Altman model for assessment of the risk of corporate bankruptcy for the Bulgarian conditions, Abstract of the dissertation, NBU-Sofia, 2017
  14. Popov D., Review of Altman's model for assessing the viability of small and medium enterprises, "Business Directions" magazine issue 2/2018
  15. Avelova D., A. Marchev, Jr. A. Marchev, Application of Z-SCORE model in business analysis, Advanced scientific tools in management, UNWE, cat. Management, Volume 1 (6) 2013, p. 261
  16. Tsvetanova E., Bankruptcy risk forecasting with a model based on discriminant analysis, adapted to the Bulgarian conditions, Godishnik of the Department of Administration and Management, NBU - Sofia, 2017

Article full text

Download PDF