Finding similar patterns in time stamped temporal datasets

Abstract

The research objective in this paper is to address the scope for research to design dissimilarity measure which uses the standard score and normal probability. Traditionally, the dissimilarity measure used is Euclidean distance to obtain dissimilarity between two known vectors of m-dimensions. The measure addressed in this paper maps the temporal pattern expressed as support vectors to z-space vectors. The dissimilarity measures uses these z-space vectors to find the dissimilarity degree between any two patterns expressed as support vectors. The procedure to retrieve similar patterns is outlined as the algorithm which uses distance and support bounds to eliminate and prune patterns that are not the required candidate patterns.