Multi-Parameter Dataset for Machine Learning Based Environmental Spoilage Risk Assessment (Cold Storage IoT)

Overview

The Multi-Parameter Dataset for Machine Learning Based Environmental Spoilage Risk Assessment published on Mendeley Data in September 2025 was compiled as part of a project to combat post-harvest food loss in developing regions through IoT-enabled cold storage monitoring.

Project Context

• Designed for smallholder farmers in Uganda to extend the shelf life of perishable agricultural products.
• Integrates IoT sensor technology with predictive machine learning to proactively control cold storage environments.
• Focuses on detecting conditions that increase spoilage risk before significant product degradation occurs.

Measured Parameters

• Internal Temperature: Cold storage chamber temperature measurements critical for food preservation.
• Internal Humidity: Relative humidity levels affecting moisture loss and microbial growth.
• Door Events: Binary or count data indicating door opening/closing frequency and duration (affecting temperature stability).
• Power Status: Electrical supply status and interruptions affecting cooling system operation.
• Ambient Conditions: External temperature and humidity for context and predictive modeling.
• Cooling System Status: Compressor on/off cycles and operational parameters.

Spoilage Risk Classification

• Data labeled with spoilage risk levels (e.g., low, medium, high) based on duration and severity of suboptimal conditions.
• Ground truth derived from food quality assessments and expert knowledge of perishable product storage requirements.
• Enables supervised learning for predictive models that classify current and forecasted environmental conditions.

Use Cases

• Predictive maintenance: Anticipating cooling system failures or power outages before food spoilage occurs.
• Smart alerts: Developing real-time notification systems for farmers when conditions exceed safe thresholds.
• Energy optimization: Balancing cooling efficiency with energy costs in off-grid or unreliable power environments.
• Food security research: Quantifying post-harvest losses and evaluating intervention effectiveness in developing regions.
• IoT for development: Designing affordable and robust cold chain monitoring solutions for smallholder agriculture.