15-minute ESP32 environmental and solar-voltage data: 84 training days and 31 validation days in Zapopan, plus a separate Ulm trial; forecasts are one step ahead.
6,461,328 annotated samples from 22 free-living participants wearing two accelerometers; CSVs contain eight core fields, with extra index columns in three files.
Anonymized elevator-door IoT sensor time series from Huawei Munich Research Center. ZIP format, 453.9 kB, sampled at 4 Hz for predictive maintenance of elevator doors.
Real train-compressor sensor dataset from Porto metro with 7,116,940 time-series instances and 21 attributes. CSV files support anomaly detection, failure prediction, and RUL research.
Synthetic steel cold-rolling predictive-maintenance benchmark with six chronological CSV streams, 51 features, and anomaly labels for work roll, bearing, motor, and reduction faults.
Comprehensive smart home dataset with 1,048,575 rows and 31 columns including timestamps, device states (TV, oven, lights, fridge) and activity labels for machine learning classification of daily activities.
A real-world, hourly dataset of electricity, heating, and cooling consumption for a commercial building, ideal for benchmarking and building energy model calibration.
Multi-resolution smart building energy dataset for forecasting competition with three versions: 1-year at 5-min intervals (v1.0), 40-day at 5-min (v2.0), and 1-day hourly (v3.x), designed to benchmark state-of-the-art energy prediction techniques.
Complete and labeled IoT dataset from physical data center with NFC smart passive temperature sensors, containing normal operations and anomalous behaviors for time-series anomaly detection in critical infrastructure environments.