CASAS ArWISE smartwatch dataset - free living, activity labels: Volume 3

Description

The Activity recognition from in-the-WIld SmartwatchEs (ArWISE) dataset is based on sensor data and activity labels collected from smart watches. Sensor data consist of 10Hz accelerometer, gyroscope, and relative location (distance from home location). Portions of the data are annotated with user-provided labels of their current activity.

Note: Other CASAS smart home and smartwatch datasets are also available, look for more at https://zenodo.org/communities/casas.

Csv files for each person’s data are organized into zip files by study collection dates. Volume 3 contains data for 43 participants collected between 2020 and 2025, divided into 2 distinct data collections. Each file includes a header with feature names. 

Features

stamp: date and time of the sensor reading (string)

yaw, pitch, roll, rotation_rate_x, rotation_rate_y, rotation_rate_z, user_acceleration_x, user_acceleration_y, user_acceleration_z: 3d movement (float)

distance_from_home: Haversine distance from the person's extracted home location (float)

course, speed, horizontal_accuracy, vertical_accuracy: float

battery_state: unplugged or charging (string)

user_activity_label: string

heart_rate: float

This dataset currently has 4 volumes:

Volume 1 doi: 10.5281/zenodo.15794726

Volume 2 doi: 10.5281/zenodo.15802788

Volume 3 doi: 10.5281/zenodo.15803341

Volume 4 doi: 10.5281/zenodo.15809156