Multivariate time-series clustering based on component relationship networks

作者：

Highlights：

• A new method transforms time series into network.

• An improved DTW calculates the distance between each component.

• The proposed clustering method with good result has little dependence on the parameter.

摘要

•A new method transforms time series into network.•An improved DTW calculates the distance between each component.•The proposed clustering method with good result has little dependence on the parameter.

论文关键词：Multivariate time-series,Clustering analysis,Component correlation,Dynamic time warping,Relational network

论文评审过程：Received 15 September 2020, Revised 11 December 2020, Accepted 20 January 2021, Available online 4 February 2021, Version of Record 24 February 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.114649