Research On Data Mining Algorithms Based On Association Rules

Data Mining is one of the most active research fields, especially in the fields of artificial intelligence and database. Data Mining is a kind of process that reveals potential useful knowledge from massive data. The association rule mining is a main research aspect of data mining. And the discovery of the frequent item sets is a key problem of the association rule mining.The most typical association rules mining algorithm is Apriori, which has been described in details in the thesis. In the process of mining frequent patterns, Apriori algorithm generates a huge number of candidate itemsets as well as needs multiple scans over database. So the time and space complexity is too high. According to the existing flaws of Apriori algorithm, we want to improve the mining efficiency from two aspects. One is to reduce the candidate itemsets and the other to decrease the scanning times over database. Firstly, an advanced algorithm is proposed to reduce the candidate itemsets depending on the nature of frequent itemsets. Then the paper propose an improved algorithm by means of coding for every item. Coding can reduce the scans over database and meanwhile deleting items can reduce the number of candidate items. As a result, the efficiency of this algorithm has been improved. Experiment done under the same conditions show that the two advanced algorithms can effectively improve the efficiency of association rules mining.