LSTPrompt: Large Language Models as Zero-Shot Time Series Forecasters by Long-Short-Term Prompting
This work addresses the challenge of improving zero-shot forecasting accuracy for time series data, which is incremental as it builds on existing prompting methods by integrating dynamic reassessment strategies.
The paper tackles the problem of zero-shot time series forecasting with large language models by proposing LSTPrompt, which decomposes forecasting into short-term and long-term sub-tasks and uses tailored prompts to enhance adaptability, resulting in consistently better performance than existing prompting methods and competitive results compared to foundation time series forecasting models.
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.