Miratar Smart Home Dataset: Motion Sensor Data from 50 Households with Synthetic Anomalies

This dataset was collected during a pilot study conducted in 50 actual homes and includes: Training dataset per user: Training data for each user, organised by specific areas of the home: Bathroom, Bedroom, Kitchen, Living room. Test dataset with anomalies per user: Test data for each user, where synthetic anomalies have been injected per area to simulate unusual activities or behaviours. Anomaly label: Ground-truth labels indicating the days on which anomalies were introduced for each user and home area. The Miratar dataset enables user-specific model training and robust evaluation in realistic smart home environments, making it a valuable resource for research on ambient-assisted living, smart home monitoring, and anomaly detection.