A tale of two toolkits, report the third: on the usage and performance of HIVE-COTE v1.0
This work provides an incremental update and practical guide for researchers and practitioners using time series classification tools.
The paper presents HIVE-COTE v1.0, an updated version of a time series classification algorithm, and evaluates its performance and resource usage compared to three recent algorithms using the aeon toolkit.
The Hierarchical Vote Collective of Transformation-based Ensembles (HIVE-COTE) is a heterogeneous meta ensemble for time series classification. Since it was first proposed in 2016, the algorithm has undergone some minor changes and there is now a configurable, scalable and easy to use version available in two open source repositories. We present an overview of the latest stable HIVE-COTE, version 1.0, and describe how it differs to the original. We provide a walkthrough guide of how to use the classifier, and conduct extensive experimental evaluation of its predictive performance and resource usage. We compare the performance of HIVE-COTE to three recently proposed algorithms using the aeon toolkit.