摘要
对于在深圳证券交易所上市的公司,通过分析和挖掘其季度报表或者相关交易网站中的数据,提取到排名预测任务中相关的数据特征以及通过爬虫获得的文本特征,成功构建了公司每股收益预测排名的模型,实现了对股价收益排名的合理预测.实验结果表明,我们的提出的模型能够有效的提高股价排名预测任务的性能,其中SPRP-Random Forests模型在NDCG@10评价指标中可以达到0.9583.在为股民选择股票,公司经营模式调整等方面具有一定的实用价值.
For companies listed on the Shenzhen Stock Exchange,by analyzing and mining the data in their quarterly reports or related transaction websites,extracting the relevant data characteristics from the ranking prediction task and the text characteristics obtained through the crawler,successfully constructed the company’s share-earnings model.The model can achieves a reasonable prediction of the ranking of valuation gains.The experimental results showthat our proposed model can effectively improve the performance of stock price ranking prediction tasks,and the SPRP-Random Forests model can reach 0.9583 in the NDCG@10 evaluation index.In the selection of stocks for investors,the company’s business model adjustment and other aspects have certain practical value.
作者
孙伯维
姚念民
孙玉轩
SUN Bo-wei;YAO Nian-min;SUN Yu-xuan(School of Computer Science,Dalian University of Technology,Dalian 116024,China)
出处
《小型微型计算机系统》
CSCD
北大核心
2020年第1期35-39,共5页
Journal of Chinese Computer Systems
基金
大连市科创基金项目(2018J12GX045)资助
关键词
排序学习
股价预测
经济数据提取
数据挖掘
单股收益排名
learning to rank
stock price forecasting
economic data extraction
data mining
single-share income ranking
