Research On Improving Algorithms Of Association Rules Mining

Recently data mining is becoming much more important as the number of databases and database size keeps growing. Researchers in many different fields have shown great interest in data mining. The scope of data mining research is also extended to various fields. Association Rules mining is one of the most hottest fields of data mining research. Data mining emerged as a rapidly growing technology in order to extract valuable information and knowledge in large volumes of data. Mining association rules is an important role of data mining because of its wide applicability in market analysis by expressing how tangible products and services relate to each other and how they tend to group together.In this paper,we first explain the issue of mining association rules is educed following the introduction of the tasks and techniques of data mining. Then, we emphasize two classical algorithms—Apriori algorithm and FP-tree algorithm and analyze some methods of improving Apriori algorithm.After analyzing the issue of mining association rules,we design two new methods of mining association rules—A method of mining association rules mining based on MAX itemsets and A method of mining association rules based on Matrix.Using MAX items to mine frequent itemsets can reduce the amount of frequent itemsets remarkably and find association rules quickly. At the same time,it avoids localization of time and space. Matrix can reduce the degree of scaning the database and improve the efficiency.Durining making association rules,it uses the knowledge of probability and reduces calculating greatly.At last,we compare the two new methods with Apriori algorithm and FP-growth algorithm by performance. And find that they are effective methods of association rulers mining.