- 无标题文档
查看论文信息

论文中文题名:

 布尔关联规则挖掘的研究及在股票价格分析中的应用    

姓名:

 孔祥纯    

学号:

 04232    

保密级别:

 公开    

学科代码:

 081203    

学科名称:

 计算机应用技术    

学生类型:

 硕士    

学位年度:

 2007    

院系:

 计算机科学与技术学院    

专业:

 计算机应用技术    

第一导师姓名:

 龙熙华    

论文外文题名:

 Boolean Mining Association Rules and its Application in Stock Price Analysis    

论文中文关键词:

 关联规则挖掘 时间序列 最大频繁项集 股票价格 股票市场    

论文外文关键词:

 Mining Association Rules Boolean Mining Association Rules    

论文中文摘要:
关联规则挖掘经过十几年的发展,取得了丰硕成果。其中的布尔关联规则挖掘是关联规则挖掘中研究比较多的一种。通过数据离散化和符号化,把时间序列数据转换为布尔型数据,从而使时间序列数据成为适合布尔关联规则挖掘的数据源。股票价格是典型的时间序列数据。随着我国股票市场进入新的发展时期,对股票市场的分析对我国的经济发展有着更加重要的意义。本文对布尔关联规则算法进行了研究,把研究成果应用在股票价格分析中。研究的工作、成果主要体现在以下三个方面: 在IODLG算法的基础上生成一种改进算法:DLG*算法。通过对IODLG算法的搜索策略进行改进,将IODLG算法与Apriori性质相结合,构造了DLG*算法。通过实验证明DLG*算法比IODLG算法更适合长频繁项集的挖掘。 在DLG算法、DMFI算法、MAFIA算法的基础上生成一种改进算法:M-DLG算法。将DLG算法的存储策略、MAFIA算法深度优先的搜索策略和DMFI算法的宽度优先的搜索策略相结合,构造了M-DLG算法。通过实验证明M-DLG算法比MAFIA算法和DMFI算法在运行效率上有了一定的提高。 把布尔关联规则挖掘应用在股票价格分析中。针对股票价格特点,开发了一个简单的股票价格分析系统原型,工作集中在数据预处理和算法实现上。以实际的股票价格数据进行了实验,实验表明该系统原型对股票价格分析是有效的。
论文外文摘要:
With a decade ofdevelopment,Mini】ng Association Rules have made much achievement. Amon8 these rules,BooleanⅣ【ining Association Rules arc mo糟frequenay studied than othel瞎.Through the Discrete and symbolized technology numerical data can be converted to Boolean data,making Time Series Data the main data SOUI'ce of Mining Association Rules. Stock price is a typical time-series data.As Chinas stock market enters a nfw era of development,the analysis of the stock market is of great significance to China's economic development.This paper conducted a study on Mining Association Rules,ap#ying research‘ results in the analysis of stock price.The effort and achievement is mainly manifested in the following three asp魄 1.DLG*Algorithm,觚improved algorithm,is generated On the basis of IODLG algorithm and the DLG algorithm.n is consU'uet司through the improvement ofthe辩缸ching strategy ofthe IODLG algorithm and the combination ofIODLG algorithm and the nature of A皿ori.Experiments have proved that DLG algorithm has n均∞advantages than IODLG algorithm in long fiequency itemset mining. 2.M-DLG Algorithm,another nfw algorithm,is generated丘om DMFI algorithm, MAFIA algorithm and the DLG algorithm.It is created through the combination of DLG Storage Strategies,the depth-first search strategy ofMAFIA algorithm and the breadth—first search swategy of DMFI.Experiments have proved that M-DLG Algorithm has been greatly improved in performance than MAFIA algorithm and DMFI algorithm. 3.Boolean M缸ng Association Rules are applied in the analysis of stock price.A simplified stock pd∞analysis prototype is created based on the previous achievements, focusing on data preprocessing and algorithm realization.Experiments have proved that this prototypeisquiteeffectiveintheanalysisofstockprice.
中图分类号:

 TP311.13 F832.5    

开放日期:

 2011-09-06    

无标题文档

   建议浏览器: 谷歌 火狐 360请用极速模式,双核浏览器请用极速模式