Time-series clustering via quasi U-statistics

Author(s)

Valk, Marcio ^[1] ; Pinheiro, Aluisio ^[2]

Affiliation

^[1] Univ Fed Rio Grande do Sul, BR-90046900 Porto Alegre, RS - Brazil

^[2] Univ Estadual Campinas, Dept Estat, Inst Matemat Estat & Comp Cient, BR-13083970 Campinas, SP - Brazil

• Document type: Journal article
• Source: JOURNAL OF TIME SERIES ANALYSIS; v. 33, n. 4, p. 608-619, JUL 2012.
• Web of Science Citations: 2

Abstract

The problem of time-series discrimination and classification is discussed. We propose a novel clustering algorithm based on a class of quasi U-statistics and subgroup decomposition tests. The decomposition may be applied to any concave time-series distance. The resulting test statistics are proven to be asymptotically normal for either i.i.d. or non-identically distributed groups of time-series under mild conditions. We illustrate its empirical performance on a simulation study and a real data analysis. The simulation setup includes stationary vs. stationary and stationary vs. non-stationary cases. The performance of the proposed method is favourably compared with some of the most common clustering measures available. (AU)

• FAPESP's process: 09/14176-8 - Quasi U-statistics, wavelets and decomposability: asymptotics and applications
• Grantee: Aluísio de Souza Pinheiro
• Support Opportunities: Regular Research Grants
• FAPESP's process: 08/51097-6 - Time Series, Dependence Analysis and Applications
• Grantee: Pedro Alberto Morettin
• Support Opportunities: Research Projects - Thematic Grants
• FAPESP's process: 07/02767-6 - Analysis of one- and multidimensional time series via quasi $U$-statistics
• Grantee: Marcio Valk
• Support Opportunities: Scholarships in Brazil - Doctorate