Are Large Language Models Useful for Time Series Data Analysis?

TL;DR

The paper investigates whether large language models (LLMs) add value for time-series data analysis by comparing LLM-based approaches to traditional non-LLM methods across classification, anomaly detection, and forecasting. It adopts a two-architecture framework: non-autoregressive LLM-enabled models for classification and anomaly detection, and autoregressive LLM-enabled models for forecasting, evaluated on multiple benchmark datasets with metrics such as accuracy, $F1$-score, $MSE$, and $MAE$. Key findings show LLMs yield gains in anomaly detection and certain classifications, but offer mixed or task-dependent advantages for forecasting, where autoregressive baselines can outperform certain LLM configurations. The results emphasize task-tailored architecture and highlight practical considerations like computational cost, providing guidance for when to deploy LLMs in time-series contexts and outlining directions for more systematic future studies.

Abstract

Time series data plays a critical role across diverse domains such as healthcare, energy, and finance, where tasks like classification, anomaly detection, and forecasting are essential for informed decision-making. Recently, large language models (LLMs) have gained prominence for their ability to handle complex data and extract meaningful insights. This study investigates whether LLMs are effective for time series data analysis by comparing their performance with non-LLM-based approaches across three tasks: classification, anomaly detection, and forecasting. Through a series of experiments using GPT4TS and autoregressive models, we evaluate their performance on benchmark datasets and assess their accuracy, precision, and ability to generalize. Our findings indicate that while LLM-based methods excel in specific tasks like anomaly detection, their benefits are less pronounced in others, such as forecasting, where simpler models sometimes perform comparably or better. This research highlights the role of LLMs in time series analysis and lays the groundwork for future studies to systematically explore their applications and limitations in handling temporal data.

Figures (2)

• Figure 1: Time Series Model with LLM (left) and without LLM (right)
• Figure 2: Left: Non-Autoregressive Model for all time series tasks GPT4TS vs. Right: Autoregressive Model autotimes for time series forecasting task