×

Time series clustering method based on centered Copula function similarity measure. (Chinese. English summary) Zbl 1474.62323

Summary: A new centered Copula function based method suitable for clustering of time series with general dependency structure is proposed. In light that the centered Copula function can capture the dependency structure between two variables, the centered Copula process is used to capture the dynamic dependency structure of the time series. Using the Cramer-von Mises statistics of the centered Copula process, a new similarity measure of time series is constructed, and a consistent nonparametric estimator and its equivalent form which is easy to calculate are given. Hierarchical clustering algorithm simulation studies show that the proposed similarity measure of time series is not only suitable for nonlinear time series data, but also has higher clustering quality for time series data with linearly dependency structures, and the practice data: GDP data in domestic provinces.

MSC:

62M10 Time series, auto-correlation, regression, etc. in statistics (GARCH)
62H30 Classification and discrimination; cluster analysis (statistical aspects)
62H05 Characterization and structure theory for multivariate probability distributions; copulas
62P20 Applications of statistics to economics
PDFBibTeX XMLCite
Full Text: DOI