Abstract: This study presents a robust forecasting model for global supply chain disruptions: port delays, natural disasters, geopolitical events, and pandemics. An integrated solution combining the help of transformer-based models for unstructured textual data preprocessing and ARIMA for structured time series analysis is referred to as a hybrid model. This model combines the insights from both approaches using a feature fusion mechanism. It evaluated the Hybrid Model using accuracy, precision, recall, and finally, F1 score, and it was found to perform much better, generally obtaining an overall accuracy of 94.2% and an overall weighted F1 score of 94.3%. Specifically, class-specific analysis demonstrated high precision in identifying disruptions such as pandemics (95.5%) and natural disasters (94.6%), showing the ability of a model to understand context and time. The proposed approach outperforms classic stand-alone statistical and deep learning models regarding scalability and adaptivity to real-life applications such as risk management and policy making. Future work could include making the weights for each cluster dynamic to optimize weights based on real-time trends and improving accuracy and resilience.

Qingzi Wang, “A Hybrid Transformer-ARIMA Model for Forecasting Global Supply Chain Disruptions Using Multimodal Data” International Journal of Advanced Computer Science and Applications(IJACSA), 16(1), 2025. http://dx.doi.org/10.14569/IJACSA.2025.0160153

@article{Wang2025,
title = {A Hybrid Transformer-ARIMA Model for Forecasting Global Supply Chain Disruptions Using Multimodal Data},
journal = {International Journal of Advanced Computer Science and Applications},
doi = {10.14569/IJACSA.2025.0160153},
url = {http://dx.doi.org/10.14569/IJACSA.2025.0160153},
year = {2025},
publisher = {The Science and Information Organization},
volume = {16},
number = {1},
author = {Qingzi Wang}
}