{Quantified Canine: Inferring Dog Personality From Wearables

Being able to assess dog personality can be used to, for example, match shelter dogs with future owners, and personalize dog activities. Such an assessment typically relies on experts or psychological scales administered to dog owners, which are costly. Instead, we built a device, called "Patchkeeper", which can be strapped on the pet's chest and measures its activity through an accelerometer and a gyroscope. After an in-the-wild deployment involving 22 dogs, we collected and analyzed both over 1300 hours of sensor data and results from two validated questionnaires capturing dog personality traits based on owners' self-reports. Using these datasets, we trained machine-learning classifiers that predicted dog personality from a combination of inertial and demographic features, achieving AUCs in [0.63-0.90]. These results suggest the value of tracking and quantifying psychological signals of pets using wearable technologies, contributing to a new research area that could be called "Pet Computing".