An Efficient Algorithm for High Utility Pattern Mining From Transactional Databases

EasyChair Preprint 2040

Abstract

Main purpose of data mining is to find  useful data set from raw data. Various data mining techniques are present. One of them is Frequent Pattern Mining technique which was used for find frequent patterns from databases. For usefulness of such frequent patterns, many constraints had been proposed by many researchers like utility parameters (price, profit, quantity etc.)as well as weight of an itemset etc. Mining high utility patterns from transaction database mainly focuses on the utility value of an itemset. Many algorithm have been proposed for finding user’s goal previously, but they contains some limitations for large datasets when number of candidates itemsets are large. And when we talk about number of itemset when large number of candidates itemsets are present as raw data, it degraded the performance of the algorithm in the  terms of memory requirement and execution time. The most significant problem of utility  mining  is  that  these  patterns  do not satisfy anti-monotonicity property and hence mining high utility patterns using traditional association rule mining algorithm becomes difficult. Additionally when long transaction are considered the situation become worse. In this paper, we present a survey and comparison among various association rule mining algorithms which deals with high utility patterns mining are considered.

Keyphrases: Assoiciation, Data Mining, Rule Mining, interestingness measure, utility mining

@booklet{EasyChair:2040,
  author    = {Ritumbara Chauhan and Kalyani Tiwari},
  title     = {An Efficient Algorithm for High Utility Pattern Mining From Transactional Databases},
  howpublished = {EasyChair Preprint 2040},
  year      = {EasyChair, 2019}}