| At information age magnanimity the appearances of data, how utilize huge the original data analysis current situation and predict futures of quantities effectively, Have already become a great "challenge that the mankind has faced. Therefore the data mining technology to be arised at the historic moment and can be developed rapidly.Data mining is a process that collectedding and implied from the database, unknown before, course with potential use value information ,It is the course that " mining knowledge" from the" voluminous" data, is a kind of new-type data analysis technology. But apply it to finance, insurance, communal facilities, government, education, telecommunication, software development of the bank, transporting,etc. on each enterprises and institutions and national defence scientific research extensively. Data mining project made organization understand better all kinds of organizations, business punish the course and customer the institutional frameworks, Thus get very high investment return .At present, the main research field of data mining is data summarizing, classifying, the cluster, Association rule,etc.. The association rule is put forward by R, Agrawal at first. The rule of a certain related relation between a group of targets in the expression database.. For example, association rules can be expressed in order to" has 95% of the persons who have bought B again in the customer of project A of purchase". Find, out customer buy behavior mode, apply to goods design, produce arrange, the marketings of pertinence,etc. by shelve from these rule. The relatively typical example is the stories of beer and diaper" to adopt the association model. It is its basic aim in order to make benefit how to guide the real work. The association ruleminings the successful application in the fields of commerce, education, scientific research etc., Which make it become a branch of most important and most active in data mining .The main research direction of association rule finds is the optimization of an algorithm and expansion application. The typical algorithms of association rule is Apriori that is put forward by R. Agrawal, minings algorithms to adopt extensively for other all kinds of cloth related rules in its key technology. However, in calculating Hou select support one who project collect, algorithm to want and scan whole data bases of circulations Apriori, But at the increases of K, not only K link figure who project collect reduce and include these things who project collected few. Because data heavy often newer scale, it makes to be relatively low to mining efficiency and design effective algorithm. And, in order to get the effective, steady, reliable association rule, should adjust support and confidence these two pieces of valve value, credibility constantly. So, it is a research subject very important how to design high-efficient renewal and maintain algorithms.The research work of this text comes from above-mentioned backgrounds . Have made following progress mainly:1. Studied algorithm of Association rule in Data Mining. Made an overall analysis of the classical Apriori algorithm and pointed out the committed steps and collected the deficiency of frequency algorithm in data mining. Have studied the current optimization algorithm situation of Apriori algorithm : On the basis of the dividing method, method based on hash, on the basis of the sampling method and so on. Studied the extension Association Rule and the technology to distill Association Rule after numerical data had been dispersed.This technology is applied to the following case study.3. studied the efficiency question of Association Rule in data mining. Put forward several directions of Association Rule optimization, such as optimize algorithm of data mining, use field knowledge, add weighing criterion, improving mining mode ,etc..4.To the deficiency of Apriori algorithm, brought forward EA, a high-efficient algorithm for mining association rule. With the result of Lkand Ck, we filter the whole database, decreas...
|
PDF Full Text Request