Frequent Data Generation Using Relative Data Analysis

Traditional association rule mining method mines association rules only for the items bought by the customer. However an actual transaction consists of the items bought by the customer along with the quantity of items bought. This paper reconsiders the traditional database by taking into account both items as well as its quantity. This new transaction database is named as bag database and each transaction consists of item along with its quantity (called itembag). This paper proposes algorithms for mining frequent items as well as rare items from the bag database. The method for mining frequent items from the database makes use of fuzzy functions to avoid sharp boundaries between itemsets and the method for mining rare items makes use of relative support to discover rare data that appear infrequently in the database but are highly associated with specific data.