Mining frequent closed itemsets using conditional frequent pattern tree

The problem of mining complete set of frequent itemsets can be reduced to the problem of mining frequent closed itemsets, which results in a much smaller set of itemsets without information loss. In this paper, we discuss an algorithm to mine frequent closed itemsets (FCI) using conditional frequent pattern tree called CFP-tree. Mining FCI from CFP-tree is simply tracing of prefix nodes, which is a straight forward method. The main advantage of this method lies in the used of simple data structures and less computations in mining FCI. From the experimental comparisons, we find that our method outperforms Apriori and Mafia and equally performs CHARM.