Fast time-series searching with scaling and shifting

Abstract

Recently, it has been found that the technique of searching for similar patterns among time series data is very important in a wide range of scientific and business applications. In this paper, we first propose a definition of similarity based on scaling and shifting transformations. Sequence A is defined to be similar to sequence B if suitable scaling and shifting transformations can be found to transform A to B. Then, we present a geometrical view of the problem so that the scaling factor and the shifting offset can be determined. Moreover, sequence searching based on tree-based indexing structure can be performed. Finally, some technical aspects are discussed and some experiments are performed on real data (stock price movement) to measure the performance of our algorithm.