Frequent pattern mining from multivariate time series data

作者：

Highlights：

• Mining frequent patterns from multivariate time series requires high memory usage.

• Multivariate time series can be converted to sequences reducing the dimensionality.

• Thus, sequential methods can be performed on transformed multivariate time series.

• The modified PrefixSpan method always outperforms Apriori based methods.

摘要

•Mining frequent patterns from multivariate time series requires high memory usage.•Multivariate time series can be converted to sequences reducing the dimensionality.•Thus, sequential methods can be performed on transformed multivariate time series.•The modified PrefixSpan method always outperforms Apriori based methods.

论文关键词：Frequent pattern mining,Knowledge discovery,Multivariate time series,Discrete sequential data,Electronic healthcare data

论文评审过程：Received 23 January 2021, Revised 28 October 2021, Accepted 19 December 2021, Available online 12 January 2022, Version of Record 29 January 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.116435