A Study on Meta Pattern Similarity of Time Series

The time series exist widely in real life domains like finance, meteorology, medical service, transportation and so on; the extraction of valuable information from time series has been recognized in recent studies. This article gives the definition of Meta pattern similarity of time series, the necessity of measuring such similarity as well as the methods of how this similarity can be measured; Based on a complete analysis of these existing methods, in characterizing the Meta pattern similarity, we may consider adding individual preference into the measure of the distance. we carried out comparative studies with them and found several drawbacks of these methods. Then we proposed a distance weighting measurement of Meta pattern similarity.