TACO: A Lightweight Tree-Based Approximate Compression Method for Time Series

André Bauer

2025

Abstract

The rapid expansion of time series data necessitates efficient compression techniques to mitigate storage and transmission challenges. Traditional compression methods offer trade-offs between exact reconstruction, compression efficiency, and computational overhead. However, many existing approaches rely on strong statistical assumptions or require computationally intensive training, limiting their practicality for large-scale applications. In this work, we introduce TACO, a lightweight tree-based approximate compression method for time series. TACO eliminates the need for training, operates without restrictive data distribution assumptions, and enables selective decompression of individual values. We evaluate TACO on five diverse datasets comprising over 170,000 time series and compare it against two state-of-the-art methods. Experimental results demonstrate that TACO achieves compression rates of up to 92%, with average compression ratios ranging from 7.55 to 20.86, while maintaining reconstruction errors as low as 10−6, outperforming state-of-the-art approaches in three of the five datasets.

Paper Citation

in Harvard Style

Bauer A. (2025). TACO: A Lightweight Tree-Based Approximate Compression Method for Time Series. In Proceedings of the 14th International Conference on Data Science, Technology and Applications - Volume 1: DATA; ISBN 978-989-758-758-0, SciTePress, pages 182-190. DOI: 10.5220/0013644600003967

in Bibtex Style

@conference{data25,
author={André Bauer},
title={TACO: A Lightweight Tree-Based Approximate Compression Method for Time Series},
booktitle={Proceedings of the 14th International Conference on Data Science, Technology and Applications - Volume 1: DATA},
year={2025},
pages={182-190},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013644600003967},
isbn={978-989-758-758-0},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 14th International Conference on Data Science, Technology and Applications - Volume 1: DATA
TI - TACO: A Lightweight Tree-Based Approximate Compression Method for Time Series
SN - 978-989-758-758-0
AU - Bauer A.
PY - 2025
SP - 182
EP - 190
DO - 10.5220/0013644600003967
PB - SciTePress