TimeStacker: A Novel Framework with Multilevel Observation for Capturing Nonstationary Patterns in Time Series Forecasting

Qinglong Liu, Cong Xu, Wenhao Jiang, Kaixuan Wang, Lin Ma, Haifeng Li
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:39929-39946, 2025.

Abstract

Real-world time series inherently exhibit significant non-stationarity, posing substantial challenges for forecasting. To address this issue, this paper proposes a novel prediction framework, TimeStacker, designed to overcome the limitations of existing models in capturing the characteristics of non-stationary signals. By employing a unique stacking mechanism, TimeStacker effectively captures global signal features while thoroughly exploring local details. Furthermore, the framework integrates a frequency-based self-attention module, significantly enhancing its feature modeling capabilities. Experimental results demonstrate that TimeStacker achieves outstanding performance across multiple real-world datasets, including those from the energy, finance, and weather domains. It not only delivers superior predictive accuracy but also exhibits remarkable advantages with fewer parameters and higher computational efficiency.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-liu25ch, title = {{T}ime{S}tacker: A Novel Framework with Multilevel Observation for Capturing Nonstationary Patterns in Time Series Forecasting}, author = {Liu, Qinglong and Xu, Cong and Jiang, Wenhao and Wang, Kaixuan and Ma, Lin and Li, Haifeng}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {39929--39946}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/liu25ch/liu25ch.pdf}, url = {https://proceedings.mlr.press/v267/liu25ch.html}, abstract = {Real-world time series inherently exhibit significant non-stationarity, posing substantial challenges for forecasting. To address this issue, this paper proposes a novel prediction framework, TimeStacker, designed to overcome the limitations of existing models in capturing the characteristics of non-stationary signals. By employing a unique stacking mechanism, TimeStacker effectively captures global signal features while thoroughly exploring local details. Furthermore, the framework integrates a frequency-based self-attention module, significantly enhancing its feature modeling capabilities. Experimental results demonstrate that TimeStacker achieves outstanding performance across multiple real-world datasets, including those from the energy, finance, and weather domains. It not only delivers superior predictive accuracy but also exhibits remarkable advantages with fewer parameters and higher computational efficiency.} }
Endnote
%0 Conference Paper %T TimeStacker: A Novel Framework with Multilevel Observation for Capturing Nonstationary Patterns in Time Series Forecasting %A Qinglong Liu %A Cong Xu %A Wenhao Jiang %A Kaixuan Wang %A Lin Ma %A Haifeng Li %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-liu25ch %I PMLR %P 39929--39946 %U https://proceedings.mlr.press/v267/liu25ch.html %V 267 %X Real-world time series inherently exhibit significant non-stationarity, posing substantial challenges for forecasting. To address this issue, this paper proposes a novel prediction framework, TimeStacker, designed to overcome the limitations of existing models in capturing the characteristics of non-stationary signals. By employing a unique stacking mechanism, TimeStacker effectively captures global signal features while thoroughly exploring local details. Furthermore, the framework integrates a frequency-based self-attention module, significantly enhancing its feature modeling capabilities. Experimental results demonstrate that TimeStacker achieves outstanding performance across multiple real-world datasets, including those from the energy, finance, and weather domains. It not only delivers superior predictive accuracy but also exhibits remarkable advantages with fewer parameters and higher computational efficiency.
APA
Liu, Q., Xu, C., Jiang, W., Wang, K., Ma, L. & Li, H.. (2025). TimeStacker: A Novel Framework with Multilevel Observation for Capturing Nonstationary Patterns in Time Series Forecasting. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:39929-39946 Available from https://proceedings.mlr.press/v267/liu25ch.html.

Related Material