摘要
本文将传统的隐马尔科夫模型中用于参数学习的Baum-Welch算法改进为变分贝叶斯算法,并将变分贝叶斯隐马尔科夫模型应用于股票价格指数预测,分别选取国外市场美股S&P500指数以及国内市场沪深300指数进行预测,并与传统的隐马尔科夫模型、BP神经网络、ARIMA模型相比较,得出结论变分贝叶斯隐马尔科夫模型对于大规模数据处理更有优势,运算速度快且预测精度更高。
In this paper, the Baum-Welch algorithm used for parameter learning in the traditional Hidden Markov Model is improved to the variational Bayes algorithm, and the variational Bayes Hidden Markov model is applied to the stock price index prediction. The S&P500 index of the United States stock in foreign markets and the HuShen300 index of the domestic market are selected for predic-tion. Compared with the traditional Hidden Markov Model, BP neural network and ARIMA model, it is concluded that the variational Bayes Hidden Markov Model has more advantages for large- scale data processing, and the operation speed is more faster and the prediction accuracy is higher.
出处
《应用数学进展》
2023年第3期1152-1163,共12页
Advances in Applied Mathematics