smartphone-and-smartwatch-activity-and-biometrics-15m6

# Smartphone and Smartwatch Activity and Biometrics 15m6 A 15.6-million-sample, multi-device time-series corpus that unites 3-axis accelerometer and gyroscope streams from 51 volunteers. Each participant carried a smartphone and smartwatch while performing 18 everyday activities for three minutes apiece, generating synchronized recordings sampled at 20 Hz. Every record is formatted as: - **subject_id** – integer 1600–1650 uniquely identifying the volunteer - **activity_code** – single ASCII letter A–S (the letter N is deliberately skipped) denoting the performed activity: - "A" - Walking - "B" - Jogging - "C" - Stairs - "D" - Sitting - "E" - Standing - "F" - Typing - "G" - Brushing Teet - "H" - Eating Soup - "I" - Eating Chips - "J" - Eating Pasta - "K" - Driking from Cup - "L" - Eating Sandwich - "M" - Kicking (Soccer Ball) - "O" - Playing Catch w/ Tennis Ball - "P" - Dribbling (Basketball) - "Q" - Writing - "R" - Clapping - "S" - Folding Clothes - **timestamp** – Unix epoch in seconds - **x, y, z** – instantaneous sensor readings for the corresponding device and modality The dataset is split into four directories—phone-accelerometer, phone-gyroscope, watch-accelerometer, watch-gyroscope—each holding 51 per-subject files. Alongside raw traces, pre-computed 10-second sliding-window features are provided, enabling both activity-recognition and behavioral-biometric research. **Acknowledgements**: The dataset is hosted at the [UCI MLR—Smartphone and Smartwatch Activity and Biometrics](https://archive.ics.uci.edu/dataset/507/wisdm+smartphone+and+smartwatch+activity+and+biometrics+dataset)