SEQUENTIAL TEMPORAL PATTERN MINING IN TIME-INTERVAL BASED EVENT DATA

Kubra Iram, T.R.Mahesh

Abstract: In one of the subfield in the world of data mining, sequential mining is one of the imperative subfield. As of late, applications utilizing time interim based occasion data have pulled in impressive endeavors in finding patterns from occasions that hold on for some span. Since the relationship between two interims is inherently unpredictable, how to viably and effectively mine interim based groupings is a testing issue. In this paper, two unusual representations, endpoint representation and endtime representation, are projected to disentangle the preparing of complex connections among occasion interims. In view of the projected representations, three sorts of interim based patterns: fleeting pattern, event probabilistic transient pattern and term probabilistic worldly pattern, are characterized. Likewise, we created two novel calculations, two algorithms mentioned below, to find three sorts of interim based successive patterns. We additionally show three pruning methods to facilitate diminish the pursuit space of the mining process. Exploratory studies demonstrate that both calculations can discover three sorts of patterns proficiently. Besides, we apply proposed calculations to genuine datasets to show the viability and approve the practicality of proposed patterns.

Keywords: Sequential Pattern Mining, Endtime, Endpoint Representations, End Time Representation.

DOI: https://doi.org/10.15623/ijret.2016.0516011