An Empirical Evaluation of Multivariate Time Series Classification with Input Transformation across Different Dimensions
This repository contains the official implementation of the experiments described in An Empirical Evaluation of Multivariate Time Series Classification with Input Transformation across Different Dimensions, accepted at ICMLA 2022 .
The code is written in Python 3.7.11 and has the following main dependencies:
- numba==0.55.1
- sympy==1.10.1
- scipy==1.4.1
- sktime==0.10.1
- sktime-dl==0.2.0
- numpy==1.21.5
- scikit-learn==1.0.2
- pandas==1.3.5
The version of the mpi4py library is 4.0.0.dev0 and has been installed from its repository.
For the WEASEL+MUSE, ResNet and InceptionTime models, MPI is used to distribute the workload of the experiments across different nodes, but no communication among nodes is necessary.
There are no separate files for the baseline experiments, but these can be derived from the existing files by skipping the data scaling alltogether.
The models are evaluated on a subset of the UEA multivariate dataset collection.
The experiment metrics can be found under Results, in the format [model_name]_uea_metrics_[scaling_method]_[dimension].csv
The baseline metrics are in the format [model_name]_uea_metrics_none.csv