Potential mapping of physical activity using machine learning technique

Authors: Kangjae Lee*, Informatics, University of Illinois at Urbana Champaign
Topics: Recreational and Sport Geography, Geographic Information Science and Systems, Transportation Geography
Keywords: physical activity, potential mapping, GIS, GPS, machine learning
Session Type: Paper
Day: 4/10/2018
Start / End Time: 12:40 PM / 2:20 PM
Room: Poydras, Sheraton, 3rd Floor
Presentation File: No File Uploaded


As one of the major issues in public health, moderate to vigorous physical activity, like jogging and brisk walking, contributes to reduced risk of physical and mental health problems. For the last decade, many scholars have attempted to examine the impact of various influential factors in surrounding environments, like built environment, on physical activity taking into account daily movement of individuals by means of GPS and accelerometer devices due to the importance of non-residential areas in people’s daily life as well as residential areas. Many existing studies have been limited to traditional approaches as to statistical analyses despite the fact that GPS trajectories have the powerful potential to shift the paradigm of public health research to data-driven analysis and modeling. Thus, data-driven approaches are required to process and create many environmental variables and construct an accurate predictive model taking into account the large size of observations and predictors. Further, the representation and exploration of physical activity patterns have not been studied well because limited areas in research regions are covered by the GPS trajectories coupled with the intensity of physical activity to investigate the association between moderate to vigorous physical activity and environmental factors. Thus, this study seeks to provide an approach to potential mapping of physical activity using machine learning techniques to explore its pattern and investigate its association with environmental contexts. This study will contribute to the introduction of the data-driven approach to physical activity research beyond the traditional statistical approaches.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login