Forecasting SCFI using a deep learning model based on Time-LLM and news data

Today, artificial intelligence (AI) is advancing rapidly and is now being utilized across various sectors of society.One of the most successful AI technologies is the Large Language Model (LLM), and recently, research on time series forecasting using LLMs has been actively conducted.Another emerging trend in time series forecasting is the development of leading indexes that can explain rapid changes in economic and social conditions in advance, using text data from news articles and social media.In this study, we propose a deep learning model that leverages LLMs and news data to forecast the Shanghai Containerized Freight Index (SCFI), addressing key recent trends in time series forecasting.SCFI reflects the volatility of container shipping rates and serves as a crucial indicator for assessing market conditions in the shipping industry.Since this index plays a significant role in the decision-making processes of logistics companies and shippers responding to shipping fluctuations, accurate forecasting of SCFI is of great importance.The proposed deep learning model demonstrates relatively higher predictive accuracy compared to competing models, even under rapidly changing SCFI conditions.As a result, it is expected to make a significant contribution to the maritime industry by enhancing decision-making processes and improving adaptability to market fluctuations.