股票时间序列模型的关联规则挖掘

Association rules mining in stock time series model

为了辅助投资者对股票投资做出正确的决策,需要对股票时间序列进行详尽的关联规则分析.而传统的关联规则只是挖掘出同一事务间的关联项,没有较高的预测价值.对此,本文建立了一种具有时间约束的关联规则挖掘模型,该模型不仅能发现股票交易同一事务间的关联项,还能挖掘出交易纪录前后的关联规则;在模型的基础上进一步给出了Optim izedApriori和ES-Apriori挖掘算法,优化了挖掘效率较低的Apriori算法.实验结果显示出在所建立的模型基础上所得的关联规则很好的反映出股票的实际情况,从而有效地为投资者服务.

In order to assist stock investors to make reasonable decision, it＇s required to lucubrate on association rules analysis. The classical researches focus on the mining of associated items within the same transaction record, thus have no forecasting value. To solve this problem, a time series model is built in this paper, which can express the associations within not only the same but also the different transaction records. Then we present two mining algorithms named Optimized Apriori and ES-Apriori that can optimize the inefficient Apriori. The model and algorithms are proved to be efficient and correct by experiments. The mining results can reflect the practical information in stock market, and can be helped to guide the investors.