Abstract
With increasing competition, the market situation is constantly changing and many enterprises are at risk of bankruptcy. There are various methods for predicting the insolvency of manufacturing enterprises, but artificial intelligence methods allow this to be more accurately. Global data used for the analysis and forecasting of bankruptcy reveal the general patterns of this economic phenomenon. An analysis of publications on predicting bankruptcy of enterprises made it possible to identify frequently used mathematical models constructed for foreign firms and giving high accuracy for Russian ones. However, a comparative analysis of various methods led to the conclusion that they need to update due to economic conditions external to the company, as well as the increased computing power of modern computers. The authors selected artificial intelligence methods that allow you to build a trained neural network and make it universal for predicting the bankruptcy of any production enterprise. The authors constructed an algorithm and a neural network, and made a bankruptcy forecast was carried out with an accuracy of 89 %. It substantiates the construction and use of a mathematical model with a high ability to predict the bankruptcy of various enterprises in any region of the world based on the latest neural network technologies of deep learning (Deep learning). Some of the deep learning technologies are the Keras and TensorFlow libraries — these are APIs (application programming interface) designed for specialists in the analysis and modeling of subject areas. The article presents the algorithm of the neural network, the results of its testing.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have