Pembuatan Modul Delete pada Aplikasi Fuzzy Temporal Association Rule Mining untuk Data Transaksi

Abstract

Transaction activities in supermarket produce large transaction data. It requires data mining techniques including association rule mining to extract patterns from the data. This research aims to implement the incremental updating technique to create association rules from expired data using fuzzy calendar on temporal database. The result is a deletion module which can find association rules without scanning the original entire database. The output are frequent itemsets and association rules in which some partitions are deleted from the original data set. The data used in this research are transaction data in a supermarket on period 1 March until 21 May 2004. The experiment was executed using support threshold values 20%, 30%, 40% and confidence threshold values 65%, 70%, 75% with early week or early year as the fuzzy calendar. By applying the deletion module the research obtains results that association rules generation are effective and efficient which means the process can produce interesting association rules in a relatively short time. For five partitions deleted data with support threshold 30% and confidence threshold 70%, one frequent itemset is generated and there is one association rule: 30(snack) → 80(milk). The execution time to generate association rules with deletion module is 13.984 seconds and the execution time without deletion module is 40.891 seconds with support threshold 40% and confidence threshold 75% based on the assumption in deletion module that frequent itemsets generated from the original data set are already provided.