PATTERN-BASED CHARACTERIZATION OF TIME SERIES
Jinfei Xie, Wenqi Yan · 2007
In this paper, two types of patterns, convex and concave patterns, are introduced to characterize finite sequences of a given time series. It is shown with an example that the pattern-based characterization of time series is promising for the purposes of subsequence matching. Statistical properties of the patterns are examined for the historical Dow Jones Index as well.