Certain investigations on mining high Utility itemsets using novel utility Pattern approaches in large Transaction databases

Abstract

Data mining is an essential technique used in the current generation for the wide applications. Technically, data mining is the process of mining the relevant patterns as useful information from the large database and solves the problems in different fields such as retail management, business, education, banking, and bioinformatics. The High Utility Itemset Mining (HUIM) is an important concept in data mining mainly used to find the highly profitable items in business application. It is a process of identifying the itemsets whose utility meets the expectation for a given threshold value. Moreover it allows user to recognize the importance or usefulness of the items by setting different threshold values. The candidates are generated ased on the given threshold value. Then High Utility Itemsets (HUI) are acquired from the possible candidate sets. But the candidate sets required more search space. The memory consumption is an important aspect to be considered during HUIM. Running time is another significant challenge to determine the UI from large search space for searching and retrieving the actual utility items among possible-high utility itemsets. Beside, more number of candidate set generation need to be reduced using advanced pruning methodologies. The process of number of tree construction and the candidate sets generation are reflected as a challenge in terms of memory and time. In this research work, techniques are proposed to mine the HUI from the large transaction database with less number of time and memory utilization. The proposed work is carried out using different approaches, namely Fuzzy Utility Pattern (FUP)-Growth newline