Enterprise Profit Forecast Model Based on Long Short-Term Memory Neural Network

摘要

In the era of the rapid development of artificial intelligence, in order to improve the usefulness of accounting information, this paper uses Long Short-Term Memory (LSTM) neural network model and financial statement information to forecast the profit of listed companies, and compares with the results predicted by analysts. In the profit forecast task of enterprises from Shanghai and Shenzhen 300 (CSI 300), the average accuracy of LSTM model is 88.6%, which is 13.52% higher than the average accuracy of analysts' forecast. In the accuracy distribution, there is no thick tail phenomenon in the results of LSTM model, and its kurtosis is significantly higher than that of analysts' forecast, and the variance is significantly lower than that of analysts' forecast. It reveals the practical significance of the application of artificial intelligence model in financial forecasting.