Mining high-utility itemset using weighted matrix-based Apriori algorithm

Abstract

Data mining is the process of discovering interesting and useful patterns and relationships in large sets of data. A recently published weighted matrix-based Apriori algorithm introduced new dimensions in the field of mining frequent patterns. The algorithm scans the transaction database and generates a 0-1 transaction matrix for getting the weighted support and confidence. The items and transactions are weighted to reflect the importance of the transaction database. The algorithm uses only minimum support as an interesting measure, often low utility items are taken as frequent itemsets, which is not beneficial for users. The field of mining high utility itemsets consider interesting utility factors like profit, cost, etc. while finding frequent patterns. This paper proposes a modified algorithm for mining high-utility itemset using a weighted matrix-based Apriori algorithm. Transaction database is transformed into a quantity transaction database. Transaction utility is calculated by multiplying each item’s internal utility with its external utility available in that transaction. The weight support and utility of items are calculated to discover high-utility frequent itemsets. The exploratory outcomes show that the modified algorithm Mining high-utility itemset using a weighted matrix-based Apriori algorithm achieves good performance in the generation of high utility frequent itemsets.