This paper analyzes the Apriori algorithm, applies hashing, transaction compression, partitioning, sampling and other methods to minimize the number of database scans, quickly find frequent itemsets, and improve the efficiency of Apriori algorithm.