An Overview on Feature-Based Classification Algorithms for Multivariate Time Series

The research on multivariate time series (MTS) has developed rapidly in the past two decades. As an important part of data mining, the classification task for MTS has gained increasing attention from experts of diverse fields. In this paper, 26 feature-based classification methods for MTS are analyzed and summarize. Since the extraction of temporal features are the core of feature-based MTS classification, these methods are mainly divided into two categories: methods with hand-crafted features and methods with learnt features. The principles and procedures of these methods are introduced, and the advantages and disadvantages are also analyzed. Besides, the recent research directions in MTS classification, such as: early classification, imbalanced classification and classification with missing value are also discussed.