Multimodal Industrial Logistics Disruption and Risk Dataset

The released dataset represents a large-scale, high-resolution real-time industrial logistics monitoring dataset, designed to support research on disruption classification, risk assessment, and resilient supply-chain analytics.It consists of hourly records collected continuously from March 14, 2021, to September 29, 2025, capturing dynamic operational, environmental, and network-level conditions across complex logistics corridors. The temporal granularity and long observation horizon enable realistic modeling of short-term fluctuations, seasonal effects, and rare disruption events. Each record corresponds to a shipment–route–facility state observed at a specific hour and integrates heterogeneous data streams commonly encountered in real-world logistics systems, including transportation management systems (TMS), warehouse management systems (WMS), IoT sensors, AIS telemetry, and external contextual sources. The dataset deliberately exhibits natural imbalance, noise, and non-stationarity, reflecting real operational environments rather than curated benchmark conditions. All identifiers are anonymized and indexed to preserve privacy while maintaining realistic correlations between shipments, routes, facilities, and zones. The dataset is intended for real-time prediction, temporal modeling, multi-class classification, robustness analysis, and explainable AI studies in industrial logistics. Feature Composition The dataset contains over 80 attributes, organized into the following categories: 1. Identification and Spatial–Temporal Context Timestamp (hourly resolution) Shipment ID, Route ID, Facility ID, Zone ID Latitude, Longitude These features define the spatio-temporal state of logistics operations. 2. Shipment and Order Characteristics Incoterm, Carrier Mode, Product Category Hazardous Material Flag Weight (kg), Volume (m³), Declared Value (USD) Priority Flag, Temperature Control Indicator Planned Transit Time, Time Buffer They capture shipment criticality, handling requirements, and economic exposure. 3. Route Execution and Reliability Metrics ETA Gap (hours), Dwell Time (minutes) Stop Count (last 24 hours) Lane Reliability Score Alternate Path Count Route Vulnerability Score These variables describe execution efficiency and routing resilience. 4. Facility and Port Conditions Yard Utilization (%), Gate Turn Time (minutes) Berth Occupancy (%) Customs Backlog Index Labor Action Flag They reflect congestion, processing delays, and workforce disruptions. 5. IoT and Telematics Signals Vehicle Speed (km/h), Heading Reefer Temperature Deviation Fuel Level (%) Engine Fault Code Count These streams model real-time asset health and mobility behavior. 6. Environmental and Geopolitical Indicators Temperature (°C), Precipitation (mm), Wind Speed (km/h) Storm Alert Level, Flood Risk Index Wildfire Risk, Earthquake Index Conflict Intensity Index Piracy Corridor Flag They enable climate- and risk-aware logistics analysis. 7. Market and Capacity Signals Spot Freight Rate (USD/TEU) Vehicle Capacity Utilization (%) Driver Availability Index Demand Index Bullwhip Index These attributes capture economic pressure and capacity imbalance. 8. Sustainability and Efficiency Metrics Planned and Actual CO₂e Emissions (kg) CO₂e Gap (%) Idle Time (minutes) Detour Distance (%) Consolidation Ratio Empty Run Ratio They support sustainability impact assessment and green logistics optimization. 9. Network and Graph Topology Features Node Degree Betweenness Centrality Community ID These features encode structural dependencies within logistics networks. 10. Calendar and Data Quality Indicators Day of Week, Hour of Day, Season Holiday Flag Data Source Type Data Completeness (%) Latency (seconds) They characterize temporal patterns and data reliability. Target Labels The dataset provides multiple unbalanced target variables for supervised learning: Disruption Type (10 classes):Normal, Weather, Port Congestion, Customs Delay, Supplier Shutdown, Labor Strike, Cyber Incident, Route Blockage, Fuel Shortage, Demand Shock Severity Level (5 levels): Ordinal scale from negligible to extreme Zone Risk Level (4 levels): Low to critical regional risk Sustainability Impact Class (4 levels): Environmental impact severity