Efficient Model Selection for Time Series Forecasting via LLMs

2025-04-02

AI Summary

Traditional time-series forecasting model selection relies on costly cross-dataset performance evaluations, while meta-learning approaches require pre-constructed, expensive performance matrices. Method: We propose the first training-free, lightweight LLM-based model selection framework for time series, eliminating the need for explicit performance matrices. Instead, it leverages the implicit domain knowledge and task-aware reasoning capabilities of large language models (e.g., LLaMA, GPT, Gemini) via prompt engineering to automatically recommend suitable forecasting models. Contribution/Results: Our approach breaks the strong dependency of meta-learning on historical evaluation data and pioneers the direct application of LLMs to time-series model selection. Extensive experiments across multiple benchmark datasets demonstrate that our method significantly outperforms conventional meta-learning and heuristic baselines in recommendation accuracy, while reducing computational overhead by orders of magnitude.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Model selection is a critical step in time series forecasting, traditionally requiring extensive performance evaluations across various datasets. Meta-learning approaches aim to automate this process, but they typically depend on pre-constructed performance matrices, which are costly to build. In this work, we propose to leverage Large Language Models (LLMs) as a lightweight alternative for model selection. Our method eliminates the need for explicit performance matrices by utilizing the inherent knowledge and reasoning capabilities of LLMs. Through extensive experiments with LLaMA, GPT and Gemini, we demonstrate that our approach outperforms traditional meta-learning techniques and heuristic baselines, while significantly reducing computational overhead. These findings underscore the potential of LLMs in efficient model selection for time series forecasting.

Problem

• Automate model selection for time series forecasting
• Reduce reliance on costly performance matrices
• Leverage LLMs for efficient computational performance

Innovation

• Leverage LLMs for model selection
• Eliminate performance matrices
• Reduce computational overhead significantly