Efisiensi Seleksi Frequent Itemsets Metode Tabel Hash Pruning pada Algoritma Apriori

Abstract

Apriori algorithm is one of the algorithms used to find frequent item sets and association rules are trying to find knowledge. Apriori will repeat the process by finding frequent itemset repeatedly in the database and ends when the candidate itemsets lifted. Apriori uses a lot of memory and the amount of execution time in finding the combination and comparison of frequent itemset. If the 100 transactions in 21m will itemsets and will read again and again, apriori less efficient when the candidate sets and frequent itemsets in large quantities. The approach taken is to make reductions candidate itemset previous generation using data structures such as hash tree data structure, the process is not much lighter and repeat the process reading the main database every time you make a comparison, counting and eliminiasi candidate support. Since all the candidates generated through transactions compared with itemsets Ck stored in the hash table buckets as Hk, the table will be read repeatedly to get candidate information until k+1 and does not reread the database. In the hash tree node that is not considered as candidates visit often and will not immediately eliminated, making the apriori process more efficient. The result is the number of candidates generated by trimming hashtree less, particularly in the C2-itemset generation and hash prion could be better in terms of execution time by up to 50%. If the hash table of data that collided in the bucket becomes a complex problem when reading data, then the solution is to use a function modulo n * n, n are the number of items on Ck-itemsets so that no collided in the bucket and the execution time for the better.