Frequent Item Set, Sequential Pattern Mining and Sequence Prediction: Structures and Algorithms

Abstract

The very basic objective of data mining and data analytics has been discovering hidden complex patterns in massive scale of data that cannot be processed by conventional statistical methodologies in the raw form. This leads to introduction of our field of research: mining of item sets and patterns from various forms of data. While talking about the versatility of data, in transaction databases, which mainly are used for worldwide E-commerce, product recommendation systems, ranking and indexing systems, search result, text analysis and recommendation system, it is a typical job for us to find the frequent, unopted and interesting item sets from the huge scale of transactions done by the customers or benefactors. This brings us to the study of frequent item set mining and recent developments of the same as high utility, fuzzy and uncertain item set mining. Where in several other kinds of researches like E-learning, bioinformatics, etc. One of the most important progressive trends is finding out patterns from sequential databases. This leads us to the adjacent topic: sequential pattern mining algorithms. The paper aims to illustrate and explicate the gradual developments on the mining of sequential patterns and transactional item sets, identify the functionalities and limitations of various algorithms and also enlighten about the recent studies and advancements in the field of research towards pattern mining and sequence prediction—which is the key point research sector of consumer behaviour analysis, stock market prediction, product tendency analysis, Genetic disease prediction, evolutionary computational biology, cognition analytics and many other most cultivated research field of today’s multidisciplinary world of science.