English

LSTPrompt: Large Language Models as Zero-Shot Time Series Forecasters by Long-Short-Term Prompting

Computation and Language 2024-02-27 v1 Artificial Intelligence

Abstract

Time-series forecasting (TSF) finds broad applications in real-world scenarios. Prompting off-the-shelf Large Language Models (LLMs) demonstrates strong zero-shot TSF capabilities while preserving computational efficiency. However, existing prompting methods oversimplify TSF as language next-token predictions, overlooking its dynamic nature and lack of integration with state-of-the-art prompt strategies such as Chain-of-Thought. Thus, we propose LSTPrompt, a novel approach for prompting LLMs in zero-shot TSF tasks. LSTPrompt decomposes TSF into short-term and long-term forecasting sub-tasks, tailoring prompts to each. LSTPrompt guides LLMs to regularly reassess forecasting mechanisms to enhance adaptability. Extensive evaluations demonstrate consistently better performance of LSTPrompt than existing prompting methods, and competitive results compared to foundation TSF models.

Keywords

Cite

@article{arxiv.2402.16132,
  title  = {LSTPrompt: Large Language Models as Zero-Shot Time Series Forecasters by Long-Short-Term Prompting},
  author = {Haoxin Liu and Zhiyuan Zhao and Jindong Wang and Harshavardhan Kamarthi and B. Aditya Prakash},
  journal= {arXiv preprint arXiv:2402.16132},
  year   = {2024}
}

Comments

9 pages, 4 figures, 3 tables, 2 page references, 2 page appendix