Forecasting Time Series with LLMs via Patch-Based Prompting and Decomposition
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909649924521984 |
|---|---|
| author | Bumb, Mayank Vemulapalli, Anshul Jella, Sri Harsha Vardhan Prasad Gupta, Anish La, An Rossi, Ryan A. Chen, Hongjie Dernoncourt, Franck Ahmed, Nesreen K. Wang, Yu |
| author_facet | Bumb, Mayank Vemulapalli, Anshul Jella, Sri Harsha Vardhan Prasad Gupta, Anish La, An Rossi, Ryan A. Chen, Hongjie Dernoncourt, Franck Ahmed, Nesreen K. Wang, Yu |
| contents | Recent advances in Large Language Models (LLMs) have demonstrated new possibilities for accurate and efficient time series analysis, but prior work often required heavy fine-tuning and/or ignored inter-series correlations. In this work, we explore simple and flexible prompt-based strategies that enable LLMs to perform time series forecasting without extensive retraining or the use of a complex external architecture. Through the exploration of specialized prompting methods that leverage time series decomposition, patch-based tokenization, and similarity-based neighbor augmentation, we find that it is possible to enhance LLM forecasting quality while maintaining simplicity and requiring minimal preprocessing of data. To this end, we propose our own method, PatchInstruct, which enables LLMs to make precise and effective predictions. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_12953 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Forecasting Time Series with LLMs via Patch-Based Prompting and Decomposition Bumb, Mayank Vemulapalli, Anshul Jella, Sri Harsha Vardhan Prasad Gupta, Anish La, An Rossi, Ryan A. Chen, Hongjie Dernoncourt, Franck Ahmed, Nesreen K. Wang, Yu Machine Learning Artificial Intelligence Computation and Language Recent advances in Large Language Models (LLMs) have demonstrated new possibilities for accurate and efficient time series analysis, but prior work often required heavy fine-tuning and/or ignored inter-series correlations. In this work, we explore simple and flexible prompt-based strategies that enable LLMs to perform time series forecasting without extensive retraining or the use of a complex external architecture. Through the exploration of specialized prompting methods that leverage time series decomposition, patch-based tokenization, and similarity-based neighbor augmentation, we find that it is possible to enhance LLM forecasting quality while maintaining simplicity and requiring minimal preprocessing of data. To this end, we propose our own method, PatchInstruct, which enables LLMs to make precise and effective predictions. |
| title | Forecasting Time Series with LLMs via Patch-Based Prompting and Decomposition |
| topic | Machine Learning Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2506.12953 |