Efficient Implementation of Leap Traversal For FP-Tree Mining

Most of the research on frequent itemset mining at early stages explored the itemset lattice breadth-wise. There after, with the advent of FP-Tree representation of transac- tional database, depth-wise exploration is found to be prom-ising. Recently COFI algorithm introduced Leap Traversal approach to extract (maximal) frequent itemsets. We propose a new array-based, data structure called Conditional Pattern Lattice (CPL) to represent the conditional pattern base suit-able for Leap Traversal. A non-recursive algorithm for imple-menting Leap Traversal using CPL is developed and its per-formance is found to be better than the other state of art min-ing methods.