Indexing of Compressed Time Series

Abstract

We describe a procedure for identifying major minima and maxima of a time series, and present two applications of this procedure. The first application is fast compression of a series, by selecting major extrema and discarding the other points. The compression algorithm runs in linear time and takes constant memory. The second application is indexing of compressed series by their major extrema, and retrieval of series similar to a given pattern. The retrieval procedure searches for the series whose compressed representation is similar to the compressed pattern. It allows the user to control the trade-off between the speed and accuracy of retrieval. We show the effectiveness of the compression and retrieval for stock charts, meteorological data, and electroencephalograms.

Compression of time series by extracting major extrema

“…We can also apply the same concept of importance levels to indexing a database of time series by their minima and maxima, which allows fast retrieval of series similar to a given pattern. We have described this indexing technique in Gandhi's master's thesis (Gandhi 2004), which is a continuation of the work by Fink and Pratt (2003) on indexing of time series. Figure 13.…”

Important extrema of time series

“…We can also apply the same concept of importance levels to index a database of time series by their minima and maxima, which allows fast retrieval of series similar to a given pattern. We have described this indexing technique in Gandhi's masters thesis [2], which is a continuation of the work by Pratt and Fink on the indexing of time series [1,8].…”

Index Financial Time Series Based on Zigzag-Perceptually Important Points

“…For querying the Landmark sequence, they recommended that the Landmark sequence must be more similar to a string sequence, rather than a multidimensional sequence; thus, string indexing is more suitable than the use of R-tree. Similar concepts regarding the identification of important points can be found in studies conducted in (Pratt and Fink, 2002;Fink and Pratt, 2003). They compressed a time series by selecting some of the minima and maxima or important points of peaks and troughs and dropping the other points.…”