The Application Of Association Rule In Stock Market Predication

With the growing range of computer applications, database technology and the continuous development of mining technology, data mining and knowledge discovery(DMKD) technology came into being, and to flourish, shown its great vitality. Data mining Association rule mining as one of the most active current research method has been widely used in Western financial industries and enterprises as well as large logistics companies and some e-commerce web site, it can be successfully predicted the banking needs of customers, forecasting such as stocks, futures and other market changes,not only bring convenience to customers but also to the enterprise has brought huge benefits.Flow of the market price of the stock reflects a barometer of economic trends, stock market research for this study therefore has national significance. In the basic premise of Basic Analysis and technical analysis the use of computer technology to technical analysis of the stock market has become a major aid, and has important significance. The main research direction of this paper is use data mining to knowledge of the stock transaction data for effective analysis and processing, to discover the intrinsic linkages between the data.Firstly, this paper make a detailed research on the data mining and particular in association rules, and then introduction the method of stock analysis and prediction of the background knowledge. And then focus on the discussion from the following three aspects.The first is the association rules algorithm, focusing on research and analysis of the apriori algorithm and the DHP algorithm, for a given support give the generation of frequent item sets, iteration and trim in a detailed description. Then analyse the efficiency of two algorithms and their advantages and disadvantages. On the basis of the two algorithms presented in the next chapter, two new algorithms E-Apriori and EH-Apriori algorithm to improve the two algorithms. On the one hand to reduce the generation of candidate item sets. On the other hand the algorithm to adapt to a broadened, to adapt them to multi-dimensional association rules.Secondly, a multi-dimensional inter-transaction association rules are proposed to adaptation the stock market characteristics. Multi-dimensional means that each collection contains a number of projects, and each project contains a number of multiple attributes. In this paper, two-dimensional database are using for example, given the base address and relative address of the project concept, to define the cross-transaction association rules, and then new algorithm to generate the biggest frequent item set,to calculate the support and trust of candidate. Then found inter-transaction association rules. In the last using experiments to test the efficiency of the algorithm.Finally, to examine the actual data analysis, and done a comparison with the actual situation, the study proved the value of doing this paper. If in a given support to get the association rules it can be used as a basis for investment.