摘要
对股票市场特征选择的相关问题进行了研究和讨论。根据震荡盒理论提出一种新的适应于与机器学习相结合的交易边界模型,通过结合基于距离的多核极限学习机(DBMK-ELM)与交易边界模型,构建基于时间序列预测的股票交易决策建议系统,使得在股票交易中能稳定获得较高的收益率并保持较低的投资风险。该系统可以快速地学习股市的历史数据,以适应快速更新的股票价格变化模式。
On issues related to the stock market feature selection were studied and discussed, a new transaction boundary model which is adapted to combine with machine learning based on the theory of shock boxes is proposed by combining multi-core Extreme Learning Machine （DBMK-ELM） based on the distance with the transaction boundary model. Construction of the suggestion system for stock transaction decision based on time series prediction enables the stock markets to get higher yields, stabilize and maintain a low investment risk. The system can quickly learn the history of stock market data in order to adapt to the rapid changes in the stock price update mode.
出处
《计算机应用与软件》
2017年第4期75-81,104,共8页
Computer Applications and Software
基金
2016年度湖南省教育厅科学研究项目(16C1658)
关键词
时间序列预测
机器学习
交易边界模型
Time series prediction
Machine learning
Transaction boundary model
