A new algorithm for discovering association rules

Efficiency is quite important for an algorithm to find frequent patterns from a large database. A new algorithm called LogECLAT algorithm which is enlightened by ECLAT algorithm uses special candidates to find frequent patterns from a continually updating database containing essential information about frequent patterns. LogECLAT algorithm can find several k-itemsets in one time of scanning database and thus the times of establishing new databases is reduced. For Apriori algorithm is widely applied to many fields, the comparison of performance is between LogECLAT algorithm and Apriori algorithm. This paper proves that LogECLAT algorithm can find frequent patterns correctly and performs better than Apriori algorithm theoretically and practically. The good performance of LogECLAT algorithm indicates that by using the special candidates can reduce the times of producing new database, and in this way efficiency of finding frequent patterns improves.