RF Deep Learning Model "Random Forest" for Forecasting Organizational Profit in the Digital Economy

Abstract

The relevance of the study lies in the fact that in modern conditions, artificial intelligence systems are increasingly used to support management decisions regarding the formation of an organization’s development strategy and other tasks. The purpose of the study is to forecast the profit of a credit institution using the Random Forest Regressor method to support management decision-making in a digital ecosystem. Among the methods in the presented work, the deep learning model "Random Forest" (RF — random forest), a multivariate regression model based on artificial intelligence, and the Graphviz program were used. Moreover, for the 7-factor DL model, the "Mean Absolute Error" (MAE) was 13.299, the "Root Mean Square Error" (MSE) reached 176.889, and the "Root Mean Square Error" (RMSE) amounted to 31289.71832. Moreover, in the 10-factor model, the above parameters had values respectively: 21.95, 20.532 and 421.563024. That is, judging by the value (MAE), the 7-factor model turned out to be more accurate. The scientific novelty lies in the fact that a machine learning model has been developed based on the "Deep Learning Random Forest" method, which provides support for management decisions, allowing one to obtain a reliable forecast of the bank’s net profit. The solution to this problem has important practical significance. Rospatent certificate for the computer software program No. 2023666284 dated July 28, 2023 was received. The hypothesis was put forward and proven that the use of the Random Forest DL artificial intelligence system makes it possible to obtain a forecast of the net profit of a credit institution. As a result of the study, conclusions were formulated. The processes of increasing digitalization are rapidly changing the financial world. An important role in modern conditions is played by the use of modern artificial intelligence systems to provide support for management decisions. The developed machine learning model based on the "Machine Learning Random Forest" method allows us to obtain a forecast of Sberbank’s net profit for the next year. Among the directions for further scientific research, it should be noted the identification of patterns and factors influencing the behavior of the resulting attribute — the net profit of a commercial bank.