Penggunaan Trie dalam Membangun Frequent ltemset Menggunakan Algoritmc FPGrowth

Abstract

FP-growth algorithm search for frequent itemset using the FP-tree data structure which is a prefix tree structure based on pattern growth algorithm. Prefix tree in FP-Growth uses the trie structure in the relationship pattern among the items in the database. Trie is a tree structure that uses strings as key, where each descendant represents a character or prefix. To find frequent itemset, FP-growth algorithm build prefix tree recursively. However, recursive function may need longer time than the Iterative (Manolopoulos 2003). Goal of this research is to create alternatives iteratively to the development of frequent iternset using trie on the FP-Growth. Research was conducted on several stages for the development of the frequent itcmset using a trie, stage through pruning and without pruning. In the last stages of testing is done using the input parameters, enter the difference in support, the number of transactions, and the items involved. Testing stage compare the execution time in the computing process through the stage with pruning and without pruning. Thing that affects the execution time is the number of items involved in development process of the frequent itemset. Frequent itemset development with the pruning process will he faster than without pruning, hecause at the time to build a combination of frequent itemset, infrequent items are not in the development process.