Optimization of Global Supply Chain Networks in Cross-Border Industrial Ecosystems

This dataset supports the multi-task machine-learning and network-design optimisation study, RAIRO – Operations Research. It models a four-echelon industrial-electronics supply chain spanning 14 national economies (Asia-Pacific, Europe and the Americas), 158 network nodes, 452 transport lanes and 72 months (January 2019 – December 2024), capturing pre-pandemic, COVID-disruption and post-disruption dynamics. The deposit comprises 14 inter-linked CSV files (≈7.3 million data cells) covering shipment flows, demand, supplier performance, inventory service levels, macroeconomic indicators, foreign-exchange rates, tariff schedules and disruption events, together with fully reproducible Python scripts for data generation, post-processing, verification and figure/table reproduction. The dataset is designed to benchmark five supervised-learning tasks (demand forecasting, lead-time regression, landed-cost regression, disruption classification and stock-out early warning) and one multi-objective network-design optimisation problem (freight cost vs. CO2 emissions vs. service level).