Research On Association Rule Mining Algorithm

Association rule algorithm is an important and practical technology in Data Mining. It is used to find out the interdependency of some data from colossal information. The main method is to scan the database and find out all of the high-frequency project teams; and utilize the high-frequency project team to differentiate all of the association rules. However, it consumes considerable calculating cost to find out the entire high-frequency project teams. To avoid this problem and improve the efficiency of association rule mining algorithm, the production of candidate data project teams and the reading times of database should be reduced effectively.This thesis puts forward a modified algorithm which is based on the analysis of interdependency mining theory and the traditional mining algorithms. It further concludes conventional property assessment method of interdependency algorithms as well. The aim of thesis is to provide foundation for the following mining algorithms property. Then this thesis analyzes Run-Length Encoding theory, through which it studies the accomplishment of interdependency rule mining problems. Based on various traditional mining algorisms, this thesis advances Run-Length Encoding mining algorithms and puts the Run-Length Encoding mining technology into practice in interdependency rule data mining.This thesis encodes transaction data into a small amount of data, and then mines the encoding data in memory directly. The method mentioned above empowers data mining the ability of capturing the inner rule of dynamic data. The merit of this algorithms lies in that it needn’t read the information in database repeatedly while the data shifts rapidly and can fresh the data in order to enhance the executive speed and improve the effectiveness of dealing with data.The association rule algorithm overcomes the shortcomings of traditional algorithms’timing data dealing. Exactly speaking, it is based on the improvement of Apriori algorithms. As the support is very low, it is possible to produce a lot of frequent set of items. In this case, this association rule algorithm is not the ideal one, but it is not discussed in this thesis.