Frequent Pattern Mining using Candidate Generation approach with Single Scan of Database

Most of the algorithms for discovering association rules require multiple passes over the database resulting in a large number of disk reads and placing a huge burden on the I/O subsystem [1]. To reduce this bottleneck in case of large databases, a new association rule mining algorithm, which uses both the Partition and the Apriori approach for calculating the frequent item sets in a single pass over the database is proposed in this paper that mainly uses two approaches: The Partition Approach, where data is mined in partitions and merges the results, and the Apriori approach that finds frequent sets within each partition. To evaluate its performance, it is compared with the existing algorithms, which require multiple database passes to generate the frequent item sets. Extensive experiments are performed and results show that time taken for the database scan is more than the time taken for candidate generation when the database size is large.