Cold-Chain Logistics IoT Dataset for Spoilage Risk and Shelf-Life Monitoring

The dataset used in this study represents a large-scale, real-world cold-chain logistics monitoring corpus collected from operational food supply networks spanning manufacturing, storage, transportation, and last-mile delivery stages. It consists of continuous, high-resolution time-series records acquired at 10-minute intervals between March 2022 and August 2025, reflecting routine industrial deployment of IoT-enabled cold-chain monitoring systems. The data were gathered from interconnected sensing devices installed across refrigerated warehouses, transport vehicles, distribution centers, retail storage units, and consumer-facing cold appliances, covering diverse logistics environments and operational conditions. Each record corresponds to a uniquely identifiable shipment instance and is linked to device-level, product-level, and route-level metadata, enabling end-to-end traceability throughout the logistics lifecycle. Feature Categories The dataset comprises several complementary feature groups: Identification and Traceability Attributes:Shipment identifiers, device IDs, product IDs, batch and lot numbers, route identifiers, geographic coordinates, and location types, supporting multi-stage logistics tracking and spatial analysis. Product and Packaging Information:Product category, packaging type, unit weight, manufacturing and expiry dates, and recommended storage temperature ranges, enabling context-aware spoilage risk assessment. Environmental and IoT Sensor Measurements:Continuous readings of temperature, humidity, CO₂ concentration, light exposure, vibration intensity, door status, power source, and cooling unit state, capturing both environmental stressors and operational anomalies. Thermal Excursion and Exposure Indicators:Out-of-range temperature flags, excursion durations, rolling temperature extrema, cumulative thermal abuse indices, and rates of temperature change, reflecting both short-term disturbances and long-term degradation effects. Logistics and Operational Dynamics:Vehicle speed, distance traveled, stop duration, dwell time, logistics status, inventory levels, and delay indicators, representing real transportation and storage behaviors. IoT Device Health and Network Quality Metrics:Sensor battery level, signal strength, packet loss rate, internal device temperature, firmware version, and recent data loss rates, supporting robustness analysis under imperfect sensing conditions. Consumer Electronics Interaction Signals:Device type, user override actions, door-open events, alert generation and response times, capturing human–system interaction effects in consumer-facing cold-chain endpoints. Target Variables The dataset supports multiple predictive tasks, including: Multi-class spoilage state classification (fresh, acceptable, degraded, spoiled), Binary spoilage detection, Remaining shelf-life estimation, and Anomaly identification under operational stress. All records were anonymized prior to release, with no personally identifiable information included.