《A Bit Level Representation for Time Series Data》.pdf

Data Mining and Knowledge Discovery, 13, 11–40, 2006 c 2006 Springer Science + Business Media, LLC. Manufactured in the United States. DOI: 10.1007/s10618-005-0028-0 A Bit Level Representation for Time Series Data Mining with Shape Based Similarity ANTHONY BAGNALL ajb@cmp.uea.ac.uk School of Computing Sciences, University of East Anglia, Norwich, UK CHOTIRAT “ANN” RATANAMAHATANA ann@cp.eng.chula.ac.th Department of Computer Engineering, Chulalongkorn University, Thailand EAMONN KEOGH eamonn@ STEFANO LONARDI stelo@ Department of Computer Science and Engineering, University of California, Riverside, USA GARETH JANACEK G.Janacek@uea.ac.uk School of Computing Sciences, University of East Anglia, Norwich, UK Received May 12, 2005; Accepted October 31, 2005 Published online: 12 May 2006 Abstract. Clipping is the process of transforming a real valued series into a sequence of bits representing whether each data is above or below the average. In this paper, we argue that clipping is a useful and flexible transformation for the exploratory analysis of large time dependent data sets. We demonstrate how time series stored as bits can be very efficiently compressed and manipulated and that, under some assumptions, the discriminatory power with clipped series is asymptotically equivalent to that achieved with the raw data. Unlike other transformations, clipped series can be compared directly to the raw data serie