Quantile cross-spectral density: A novel and effective tool for clustering multivariate time series
作者：
Highlights：
• A new measure based on quantiles to perform clustering of multivariate time series.
• The measure examines simultaneously both cross-dependence and serial dependence.
• The proposed measure takes advantage of the nice properties of the quantiles.
• Accurate, robust and efficient clustering performance in the frequency domain.
• The methodology is successfully applied to cluster time series of S&P 500.
摘要
•A new measure based on quantiles to perform clustering of multivariate time series.•The measure examines simultaneously both cross-dependence and serial dependence.•The proposed measure takes advantage of the nice properties of the quantiles.•Accurate, robust and efficient clustering performance in the frequency domain.•The methodology is successfully applied to cluster time series of S&P 500.
论文关键词：Multivariate time series,Clustering,Dissimilarity measure,Quantile cross-spectral density,S&P 500,UEA archive
论文评审过程：Received 4 November 2020, Revised 15 July 2021, Accepted 25 July 2021, Available online 29 July 2021, Version of Record 10 August 2021.
论文官网地址：https://doi.org/10.1016/j.eswa.2021.115677
