STOCK MARKET FORECASTING BASED ON WAVELET AND LEAST SQUARES SUPPORT VECTOR MACHINE

摘要：

In this paper, we propose a novel method using wavelet transform to denoise the input of least squares support vector machine for classification of closing price of stocks. The proposed method classifies closing price as either down or up. We have tested the proposed approach using passed three-year data of 10 stocks randomly selected from sample stock of hs300 index and compared the proposed method with other machine learning methods. Good classification percentage of almost 99％ was achieved by WT-SVM model. We observed that the performance of stock price prediction can be significantly enhanced by using hybrized WT in comparison with a single model.

关键词： Wavelet transforms Least squares support vector machine Stock price prediction Noisy signal Machine learning

作者: Xia Liang Haoran Zhu Xun Liang

作者单位: School of Information, Renmin University of China, Beijing, China

会议类型: 国际会议

会议名称: 13th International Conference on Enterprise Information System(第13届企业信息系统国际会议 ICEIS 2011)

会议地点: 北京

会议语种:英文

页码: 2515-2522

在线出版日期: 2011-06-08（万方平台首次上网日期，不代表论文的发表时间）