Stock Time Series Prediction Based on Deep Learning
Download as PDF
Zou Cunzhu, Luo Jiping, Bai Shengyuan, Wang Yuanze, Zhong Changfa, Cai Yi
With the continuous development of financial markets and the gradual improvement of the financial system, people participate in financial market investment. The interest in capital is also growing, and it is accompanied by a strong demand for accurate and effective financial information services. So how to accurately predict the trend of stocks has become a focus of attention. In this paper, based on the traditional method ARIMA, the corresponding RNN (LSTM) model is proposed for the stock time series prediction problem, and its application situation is further analyzed and optimized, so that it can better explore the change law of stock data. And by setting the corresponding experimental test model method on the stock forecasting task performance. The research and evaluation of the model method demonstrates the good performance of the deep learning model and the ARIMA model in the stock time series forecasting task. The error between the stock forecasting result and the real value of each model method is at a low level. In comparison with the prediction effects of model methods such as Prophet, the RNN model proposed in this paper is closer to the real market performance, and has achieved a significantly better prediction effect than the comparison method.
Rnn, Time Series, Arima, Stock