RealLife Activity & Environment Dataset (RAED)

This dataset presents a large-scale collection of human behavioural patterns and corresponding environmental conditions recorded across various real-world indoor and outdoor contexts.
It captures the interaction between daily human activities and environmental parameters such as temperature, humidity, noise, light, and motion intensity.
The dataset is designed to reflect diverse natural living conditions — from quiet study environments to noisy outdoor and stressful scenarios — making it valuable for research on behaviour recognition, IoT-based smart environments, context-aware systems, and human–machine interaction studies.
Each record includes a timestamp, location tag, and the observed human behaviour, alongside multiple environmental readings.
To maintain authenticity, the data includes realistic fluctuations, time irregularities, and occasional missing readings similar to those seen in actual sensor-collected datasets.

Key Features
1. Contains 30,000+ records covering human activities across diverse real-life environments.
2. Includes five behaviour categories: sleeping, studying, walking, eating, and stressed.
3. Provides environmental correlations (temperature, humidity, noise, light, motion).
4. Realistic missing values, time irregularities, and contextual variations to mimic real-world data collection.
5. Balanced yet naturally distributed data for each behaviour type.

Advantages
1. Highly natural and realistic — closely represents genuine human-environment interactions.
2. Noise, missing values, and randomness emulate practical sensor data challenges.
3. Suitable for both supervised and unsupervised learning research.
4. Captures multi-dimensional relationships between behaviour and environment.
5. Flexible and easy-to-integrate for IoT, ML, and HCI experiments.

Limitations
1. Does not include direct physiological signals (e.g., heart rate, ECG).
2. Behaviour labels are contextual and may overlap (e.g., “studying” in “Cafe”).
3. Limited to five behaviour types; future work may include more granular actions.
4. Environmental readings are derived approximations, not from physical sensors.

Potential Applications
1. Human behaviour recognition and activity classification
2. IoT-based smart environment adaptation systems
2. Ambient intelligence and context-aware computing
3. Smart home and workplace automation research
4. Human–environment interaction analysis
5. Behaviour prediction under environmental stress factors