Survey on high utility oriented sequential pattern mining

Abstract

Main purpose of using different data mining techniques is to find novel, potentially useful patterns which can be useful in real world applications to derive best knowledge and using it into useful way. For identifying the usefulness of patterns, many constraints has been proposed by many researchers are like time-interval between itemsets, utility parameters (price, profit, quantity etc.), weight of an itemset etc. Among these, time-interval between items is useful to learn about at what time next item will be purchased and utility parameters are useful to mine patterns which are more profitable in real world from huge transaction database. Here, we miss hybridization of these constraints so that more useful patterns can be derived. Hybridization of these constraints can give more useful pattern by combining advantages of each constraint. Organization of this survey is given as follows: In first section we introduced basic terms like Data mining, Frequent pattern mining, Sequential pattern mining, Time-interval Sequential Pattern Mining and utility mining. In second section we summarizes some important previous research work related to Sequential pattern mining, Time-interval based sequential pattern mining and utility mining. In last section we concluded the survey work by suggesting some future directions for discovering some strong sequential rules that can infer from extracted patterns.