A SPEED IMPROVEMENT SCHEME IN KNOWLEDGE DISCOVERY OF ASSOCIATION RULE ALGORITHM FOR A MARKET BASKET

摘要

The paper presents a scheme for improving a speed of discovery the associate rules for a market basket. The aim is to reduce the number of passes over the database. It consists of two steps, pre-processing and pattern classification. In pre-processing process, all transactions are sorted by using the number of items in each transaction. Then, in pattern classification process, all groups of items, which frequently appear together in transaction, are classified. The algorithm is implemented and tested using varieties of minimum support values and sizes of standard synthetic data set. The experimental results are compared with that of the Apriori and DIC Algorithm. The results show that the execution time of the proposed scheme is less than that of the Apriori and DIC algorithm.