Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Time Series Embedding Methods for Classification Tasks: A Review

Habib Irani, Yasamin Ghahremani, Arshia Kermani, Vangelis Metsis

TL;DR

This paper surveys time series embedding methods by organizing them into a taxonomy spanning statistical, transformation-based, feature-based, model-based, kernel-based, graph-based, manifold learning, topological, deep learning, and hybrid approaches. It conducts a rigorous, open-source, real-world dataset–driven evaluation of embeddings as fixed-dimensional representations for downstream classification, highlighting strong performance from FFT, Wavelet, and PCA across domains, while deep learning-based embeddings require substantial data and tuning. Key contributions include a systematic comparison across 11 datasets, detailed methodological guidelines, and an open-source suite enabling researchers to reproduce and extend the evaluation. The findings emphasize dataset-dependent success and the value of aligning embedding choice with signal characteristics and downstream classifiers, offering practical guidance for practitioners and a foundation for future hybrid and interpretable embedding methods.

Abstract

Time series analysis has become crucial in various fields, from engineering and finance to healthcare and social sciences. Due to their multidimensional nature, time series often need to be embedded into a fixed-dimensional feature space to enable processing with various machine learning algorithms. In this paper, we present a comprehensive review and quantitative evaluation of time series embedding methods for effective representations in machine learning and deep learning models. We introduce a taxonomy of embedding techniques, categorizing them based on their theoretical foundations and application contexts. Our work provides a quantitative evaluation of representative methods from each category by assessing their performance on downstream classification tasks across diverse real-world datasets. Our experimental results demonstrate that the performance of embedding methods varies significantly depending on the dataset and classification algorithm used, highlighting the importance of careful model selection and extensive experimentation for specific applications. To facilitate further research and practical applications, we provide an open-source code repository implementing these embedding methods. This study contributes to the field by offering a systematic comparison of time series embedding techniques, guiding practitioners in selecting appropriate methods for their specific applications, and providing a foundation for future advancements in time series analysis.

Time Series Embedding Methods for Classification Tasks: A Review

TL;DR

This paper surveys time series embedding methods by organizing them into a taxonomy spanning statistical, transformation-based, feature-based, model-based, kernel-based, graph-based, manifold learning, topological, deep learning, and hybrid approaches. It conducts a rigorous, open-source, real-world dataset–driven evaluation of embeddings as fixed-dimensional representations for downstream classification, highlighting strong performance from FFT, Wavelet, and PCA across domains, while deep learning-based embeddings require substantial data and tuning. Key contributions include a systematic comparison across 11 datasets, detailed methodological guidelines, and an open-source suite enabling researchers to reproduce and extend the evaluation. The findings emphasize dataset-dependent success and the value of aligning embedding choice with signal characteristics and downstream classifiers, offering practical guidance for practitioners and a foundation for future hybrid and interpretable embedding methods.

Abstract

Time series analysis has become crucial in various fields, from engineering and finance to healthcare and social sciences. Due to their multidimensional nature, time series often need to be embedded into a fixed-dimensional feature space to enable processing with various machine learning algorithms. In this paper, we present a comprehensive review and quantitative evaluation of time series embedding methods for effective representations in machine learning and deep learning models. We introduce a taxonomy of embedding techniques, categorizing them based on their theoretical foundations and application contexts. Our work provides a quantitative evaluation of representative methods from each category by assessing their performance on downstream classification tasks across diverse real-world datasets. Our experimental results demonstrate that the performance of embedding methods varies significantly depending on the dataset and classification algorithm used, highlighting the importance of careful model selection and extensive experimentation for specific applications. To facilitate further research and practical applications, we provide an open-source code repository implementing these embedding methods. This study contributes to the field by offering a systematic comparison of time series embedding techniques, guiding practitioners in selecting appropriate methods for their specific applications, and providing a foundation for future advancements in time series analysis.
Paper Structure (49 sections, 15 equations, 3 figures, 5 tables)

This paper contains 49 sections, 15 equations, 3 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Statistical Methods
  4. Transformation-Based Methods
  5. Feature-Based Methods
  6. Model-Based Methods
  7. Kernel-Based Methods
  8. Graph-Based Methods
  9. Manifold Learning and Nonlinear Dimensionality Reduction
  10. Topological Methods
  11. Deep Learning-Based Methods
  12. Hybrid Methods
  13. Evaluation Methodology
  14. Data
  15. Machine Learning Pipeline
  16. ...and 34 more sections

Figures (3)

  • Figure 1: Taxonomy of Time Series Embedding Methods
  • Figure 2: Machine learning pipeline for time series classification.
  • Figure 3: UMAP projections of different embedding methods on the UniMiB dataset.