Author(s)

Zhiyuan Ma, Zeyu Zhang, Xue Zhu

Corresponding Author

Xue Zhu

ABSTRACT

Today, stock investing is increasingly becoming one of the main ways people invest in finance. Therefore, reasonable short-term stock price forecasts are of great significance to investment decisions. This paper combines radial basis functions (RBFs) and support vector machines (SVMs) with bagging ensemble learning (EL) to form a new ensemble model for predicting short-term stock prices from stock market indicators. This article considers the research object of the Shanghai Stock Exchange Index closing price from January 4, 2021, to January 31, 2021. The conclusions are as follows: (1) The Bagging ensemble learning model based on RBF neural network and SVM support vector machine has a correlation coefficient R2 closer to 1 and a small mean absolute error MAE and root mean square error compared with a single model. The fitting effect is good performance, the relative error of the prediction result is low, and it is more suitable for predicting short-term stock prices. (2) In short-term stock price forecasting, the SVM support vector machine has a better general forecasting effect than RBF neural network, its correlation coefficient R2 is closer to 1, its mean absolute error MAE and root mean square error RMSE is smaller, and the prediction results of the RBF neural network are unstable.

KEYWORDS

Machine learning, RBF, SVM, Bagging ensemble learning, short-term stock price forecast