Title: Activity Classification for User’s Activities of Daily Living
Date/Time: April 13th, 2017 @ 3:00 PM
Place: 223 Atanasoff
Major Professor: Carl Chang
Committee Members: Simanta Mitra, Johnny Wong
In the era of the Internet of Things, our environment is surrounded by all kinds of sensors. These sensors are almost ubiquitous y embedded such as on airplanes and in the smartphones. Many researchers use sensors to detect and recognize our activities of daily living(ADL). Feng et al  developed an ADL Recorder Android App and ADL recognition system to detect users’ daily activities. I studied the effectiveness of ADL recognition system in recognizing and classifying activities based on the sensor data from the ADL Recorder Android App. By using the ADL Recorder Android App, I collected three weeks’ activity data from a subject. Sensors in that app record 21 attributes, such as, light, sound, latitude, longitude, and so on. We tried several classification approaches, such as, Random Forest, J48. The overall accuracy of activity classification could be as high as 88%. We also developed a web application to visualize data from the ADL recognition system so that the subject can understand his/her daily activities better via good data visualization. To verify that our web application can indeed help the subject understand his/her daily activities, I did a survey on a group of students at Iowa State University(ISU). Based on the survey data, although subjective, we conclude that potential users can use our web application to collect and visualize their ADL data and perhaps adjust their daily activities for a healthier lifestyle.
 Feng, Y., Chang, C. K., & Chang, H. (2016, May). An ADL Recognition System on Smart Phone. In Proceedings of the 2016 International Conference on Smart Homes and Health Telematics, pp. 148-158.