Temporal Fusion Transformers For High-Accuracy Forecasting Of Global Raw Material Trade

Matteo Bianchi; Dr. Lina Philip

Authors

Matteo Bianchi Phd Student at Department of Data Science and Economics, Bocconi University, Milan, Italy
Dr. Lina Philip Center for Trade Analytics, Qatar Computing Research Institute, Doha, Qatar

Keywords:

International Trade Forecasting, Raw Materials, Temporal Fusion Transformer, Time Series Analysis

Abstract

Accurate forecasting of international raw material trade flows is critical for effective policy-making, strategic supply chain management, and mitigating risks in an increasingly volatile global economy characterized by polycrises and supply chain disruptions. Traditional forecasting methods, while valuable, often struggle to capture the complex temporal dynamics and interplay of diverse influencing factors. This article explores the application of the Temporal Fusion Transformer (TFT), a state-of-the-art deep learning model, for achieving high-accuracy predictions of international raw material trade flows. We outline a conceptual framework for utilizing the TFT, highlighting its ability to leverage multiple time series inputs, incorporate static and dynamic exogenous variables, and provide interpretable insights into the drivers of trade. By comparing its potential performance against established models like ARIMA, Prophet, LSTM, and Graph Neural Networks (GNNs), we demonstrate the theoretical advantages of the TFT for this challenging forecasting task. The discussion emphasizes the implications of improved forecasting accuracy for enhancing resilience in global value chains and navigating turbulent times. While acknowledging data requirements and model complexity, this article posits that the Temporal Fusion Transformer represents a significant advancement in the toolkit for predicting international raw material trade, offering both enhanced accuracy and crucial interpretability.

References

Aamir S, Mughal SF, Kayani AJ, Yousuf MZ, Rastgar OA, Syed AA (2024). Impact of generative AI in revolutionizing education. In: 2024 8th International Symposium on Multidisciplinary Studies and Innovative Technologies (ISMSIT), 1–6.

Saboia, J.: Autoregressive integrated moving average (ARIMA) models for birth forecasting. J. Am. Stat. Assoc. 72(358), 264–270 (1977).

Sean, J.: Taylor and Benjamin Letham. Forecasting at scale. Am. Stat. 72(1), 37–45 (2018).

Tiits, M., Kalvet, T., Ounoughi, C., Yahia, S.B.: Relatedness and product complexity meet gravity models of international trade. J. Open Innov.: Technol. Mark. Complex. 10(2), 100288 (2024).

Shen, M.-L., Lee, C.-F., Liu, H.-H., Chang, P.-Y., Yang, C.-H.: Effective multinational trade forecasting using LSTM recurrent neural network. Expert Syst. Appl. 182, 115199 (2021).

Rincon-Yanez, D., Ounoughi, C., Sellami, B., Kalvet, T., Tiits, M., Senatore, S., Yahia, S.B.: Accurate prediction of international trade flows: leveraging knowledge graphs and their embeddings. J. King Saud Univ. Comput. Inf. Sci. 35(10), 101789 (2023).

Binrong, W., Wang, L.: Two-stage decomposition and temporal fusion transformers for interpretable wind speed forecasting. Energy 288, 129728 (2024).

Binrong, W., Wang, L., Zeng, Y.-R.: Interpretable tourism demand forecasting with temporal fusion transformers amid covid-19. Appl. Intell. 53(11), 14493–14514 (2023).

Jicha, F., Hassink, R., Chu, H.: Advancing regional economic resilience theorizing by integrating global value chain resilience. In: The Changing Economic Geography of Companies and Regions in Times of Risk, Uncertainty, and Crisis, pp. 32–46. Routledge (2024).

Lawrence, M., Homer-Dixon, T., Janzwood, S., Rockstörm, J., Renn, O., Donges, J.F.: Global polycrisis: the causal mechanisms of crisis entanglement. Global Sustain. 7, 6 (2024).

Davies, M., Hobson, C.: An embarrassment of changes: international relations and the covid-19 pandemic. Austral. J. Int. Aff. 77(2), 150–168 (2023).

United Nations Development Programme Regional Bureau for Asia and the Pacific. Polycrisis and long-term thinking. Technical report, United Nations Development Programme (2022).

World Economic Forum: The global risks report 2023. Technical report, World Economic Forum, Geneva (2023).

Neise, T., Verfürth, P., Franz, M.: The Changing Economic Geography of Companies and Regions in Times of Risk, Uncertainty, and Crisis. Routledge (2024).

Baldwin, R., Freeman, R.: Risks and global supply chains: what we know and what we need to know. Annu. Rev. Econ. 14, 153–180 (2022).

Xing, Y., Wang, R., Dollar, D.: Global value chain development report 2023: Resilient and sustainable GVCS in turbulent times. Technical report, World Trade Organization (2023).

OECD. OECD inventory of export restrictions on industrial raw materials 2024: monitoring the use of export restrictions amid market and policy tensions. Technical report, OECD Publishing, Paris, (2024).

Georgitzikis, K.: Trade codes of non-food, non-fuel raw materials and their products. (JRC136041) (2023).

Schimpf, S., Sturm, F., Correa, V., Bodo, B., Keane, C.: The world of raw materials 2050: scoping future dynamics in raw materials through scenarios. Energy Procedia, 125:6–13 (2017). European Geosciences Union General Assembly 2017, EGU Division Energy, Resources & Environment (ERE).

Nassar, N.T., Brainard, J., Gulley, A., Manley, R., Matos, G., Lederer, G., Bird, L.R., Pineault, D., Alonso, E., Gambogi, J., Fortier, S.M.: Evaluating the mineral commodity supply risk of the U.S. manufacturing sector. Sci. Adv. 6(8), eaay8647 (2020).

Carrara, S., Bobba, S., Blagoeva, D., Alves Dias, P., Cavalli, A., Georgitzikis, K., Grohol, M., Itul, A., Kuzov, T., Latunussa, C., Lyons, L., Malano, G., Maury, T., Prior Arce, A., Somers, J., Telsnig, T., Veeh, C., Wittmer, D., Black, C., Pennington, D., Christou, M.: Supply chain analysis and material demand forecast in strategic technologies and sectors in the eu - a foresight study. Policy assessment, Anticipation and foresight, Risk assessment KJ-NA-31-437-EN-N (online), KJ-NA-31-437-EN-C (print), Luxembourg (Luxembourg) (2023).

Tiits, M., Karo, E., Kalvet, T.: Small countries facing the technological revolution: fostering synergies between economic complexity and foresight research. Compet. Rev. 35(1), 53–75 (2025).

Capoani, L.: Review of the gravity model: origins and critical analysis of its theoretical development. SN Bus. Econ. 3(5), 95 (2023).

Sharma, P., Rohatgi, S., Jasuja, D.: Scientific mapping of gravity model of international trade literature: a bibliometric analysis. J. Scientometr. Res. 11(3), 447–57 (2023).

Baldwin, R., Taglioni, D.: Gravity for dummies and dummies for gravity equations. NBER Working Paper 12516 (2006).

Yotov, Y.V., Piermartini, R., Larch, M. et al.: An advanced guide to trade policy analysis: the structural gravity model. WTO iLibrary (2016).

Box, G., Pierce, D.A.: Distribution of residual autocorrelations in autoregressive-integrated moving average time series models. J. Am. Stat. Assoc. 65(332), 1509–1526 (1970).

Sean, J.: Taylor and Benjamin Letham. Forecasting at scale. PeerJ 5, e3190v2 (2017).

Yenidoğan, I., Çayir, A., Kozan, O., Dağ, T., Arslan, C.: Bitcoin forecasting using ARIMA and prophet. In: 2018 3rd International Conference on Computer Science and Engineering (UBMK). IEEE, pp. 621–624 (2018).

Habib Ahmed Elsayir: Forecasting ARIMA model for foreign trade statistics. Stand. Sci. Res. Essays 2(12), 636–648 (2014).

Sharma, P.: Extrapolative modeling of international trade using ARIMA model: a performance comparison between India and China during covid-19. SAMVAD: SIBM Pune Res. J. XXI, 71–91 (2020).

Aliul Islam, Md., Sumy, M., Uddin, M.A., Hossain, M.S.: Fitting ARIMA model and forecasting for the tea production, and internal consumption of tea (per year) and export of tea. Int. J. Mater. Math. Sci. 2(1), 8–15 (2020).

Güleryüz, D., Özden, E.: The prediction of brent crude oil trend using LSTM and facebook prophet. Avrupa Bilim ve Teknoloji Dergisi 20, 1–9 (2020).

Aytaç, E.: Forecasting turkey’s hazelnut export quantities with facebook’s prophet algorithm and box-cox transformation. ADCAIJ: Adv. Distrib. Comput. Artif. Intell. J. 10(1), 33 (2021).

Sellami, B., Ounoughi, C., Kalvet, T., Tiits, M., Rincon-Yanez, D.: Harnessing graph neural networks to predict international trade flows. Big Data Cogn. Comput. 8(6), 65 (2024).

Minakawa, N., Izumi, K., Sakaji, H.: Bilateral trade flow prediction by gravity-informed graph auto-encoder. In: 2022 IEEE International Conference on Big Data (Big Data). IEEE, pp. 2327–2332 (2022).

Monken, A., Haberkorn, F., Gopinath, M., Freeman, L., Batarseh, F.: Graph neural networks for modeling causality in international trade 34, 04 (2021).

Ahmed, F., Cui, Y., Fu, Y., Chen, E.: A graph neural network approach for product relationship prediction. In: International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, vol. 85383, pp. V03AT03A036. American Society of Mechanical Engineers (2021).

Panford-Quainoo, K.,Bose, A.J., Defferrard, M.: Bilateral trade modelling with graph neural networks. In: ICLR Workshop on Practical ML for Developing Countries (2020).

Verstyuk, S., Douglas, M.R.: Machine learning the gravity equation for international trade. Available at SSRN 4053795 (2022).

Yang, C.-H., Lee, C.-F., Chang, P.-Y.: Export-and import-based economic models for predicting global trade using deep learning. Expert Syst. Appl. 218, 119590 (2023).

Liu, H., Dong, Z., Jiang, R., Deng, J., Deng, J., Chen, Q., Song, X.:. Spatio-temporal adaptive embedding makes vanilla transformer sota for traffic forecasting. In: Proceedings of the 32nd ACM International Conference on Information and Knowledge Management, pp. 4125–4129 (2023).

Li, F., Wan, Z., Koch, T., Zan, G., Li, M., Zheng, Z., Liang, B.: Improving the accuracy of multi-step prediction of building energy consumption based on EEMD-PSO-Informer and long-time series. Comput. Electr. Eng. 110, 108845 (2023).

Wang, Z., Chen, Z., Yang, Y., Liu, C., Li, X., Jinran, W.: A hybrid autoformer framework for electricity demand forecasting. Energy Rep. 9, 3800–3812 (2023).

Lim, B., Arık, S.Ö., Loeff, N., Pfister, T.: Temporal fusion transformers for interpretable multi-horizon time series forecasting. Int. J. Forecast. 37(4), 1748–1764 (2021).

Binrong, W., Wang, L., Tao, R., Zeng, Y.-R.: Interpretable tourism volume forecasting with multivariate time series under the impact of covid-19. Neural Comput. Appl. 35(7), 5437–5463 (2023).

Han, Y., Tian, Y., Liangliang, Yu., Gao, Y.: Economic system forecasting based on temporal fusion transformers: multi-dimensional evaluation and cross-model comparative analysis. Neurocomputing 552, 126500 (2023).

Breiman, L.: Random forests. Mach. Learn. 45, 5–32 (2001).

Hochreiter, S.: Long Short-Term Memory. Neural Computation MIT-Press (1997).

Scarselli, F., Marco Gori, A., Tsoi, C., Hagenbuchner, M., Monfardini, G.: The graph neural network model. IEEE Trans. Neural Netw. 20(1), 61–80 (2008).

Zhou, H., Zhang, S., Peng, J., Zhang, S., Li, J., Xiong, H., Zhang, W.: Informer: beyond efficient transformer for long sequence time-series forecasting. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 35, pp. 11106–11115 (2021).

Haixu, W., Jiehui, X., Wang, J., Long, M.: Autoformer: decomposition transformers with auto-correlation for long-term series forecasting. Adv. Neural Inf. Process. Syst. 34, 22419–22430 (2021).

Li, S., Jin, X., Xuan, Y., Zhou, X., Chen, W., Wang, Y.-X., Yan, X.: Enhancing the locality and breaking the memory bottleneck of transformer on time series forecasting. Advances in neural information processing systems, 32 (2019).

Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, Ł., Polosukhin, I.: Attention is all you need. Advances in neural information processing systems, 30 (2017).

Kattel, R., Randma-Liiv, T., Kalvet, T.: Small states, innovation and administrative capacity. In: Bekkers, V., Edelenbos, J., Steijn, B. (eds.) Innovation in the Public Sector: Linking Capacity and Leadership, pp. 61–81. Palgrave Macmillan UK, London (2011).

Tiits, M., Kalvet, T.: Nordic Small Countries in the Global High-tech Value Chains: The Case of Telecommunications Systems Production in Estonia. Technical report, Working Papers in Technology Governance and Economic Dynamics (2012).

Lai, Z., Tangjie, W., Fei, X., Ling, Q.: Bert4st: fine-tuning pre-trained large language model for wind power forecasting. Energy Convers. Manag. 307, 118331 (2024).

Xue, H., Salim, F.D.: Promptcast: a new prompt-based learning paradigm for time series forecasting. IEEE Trans. Knowl. Data Eng. 36, 6851 (2023).

Frontiers in Emerging Artificial Intelligence and Machine Learning

Article Details Page