Activity recognition within smart homes using logistic regression

Abstract

This study takes a look and investigates the application of logistical regression to recognize multiple predefined human activities performed by a single test subject within smart home environment. Two Inertial Measurement Unit (IMU) devices were used as wearable gadgets to collect movement data from the test subjects. Test subject worn these devices on their right wrist and ankle. The obtained data are combined room air quality data (Humidity, CO2, temperature) measured using KNX smart home technology. The research is beneficial to provide detailed report about activity levels and well-being of the smart home residence and create a smart home environment that adjust automatically to its residence requirements. The obtained results showed very high classification accuracy (97.8% in average).