| Title |
Comparative Study on the Short-term Forecasting Performance of LLM Models for Time Series Data |
| Authors |
정현욱(Hyunwook Jeong) ; 정은성(Eun-Sung Jung) |
| DOI |
https://doi.org/10.5573/ieie.2025.62.6.29 |
| Keywords |
Time series forecasting; Large language model; Short-term forecasting; Patching; Transformer |
| Abstract |
Since the Transformer model was introduced, numerous time series forecasting methods have been proposed based on its architecture. This paper introduces various methods of time series forecasting and presents representative studies for each method. Also, we evaluate the short-term forecasting performance of Time-LLM, which is known to be the most advanced model for long-term time series forecasting among recent related time series forecasting studies. Current research on the applicability of Time-LLM has primarily focused on long-term forecasting, while studies on short-term forecasting remain relatively limited. So, we compare and analyze the short-term time series forecasting performance of Time-LLM against other models using datasets with various distributions. We also compare and analyze the short-term time series forecasting performance based on the performance of the backbone models. |