Analyzing the Mobility of University Members for InnaMoRuhr

The goal of the InnaMoRuhr research project is the development of an integrated and sustainable mobility concept for the University Alliance Ruhr. As basis for this concept, we have been studying the mobility of the university staff and students during field trials. To capture the mobility choices of the participants, we have developed and deployed mobile applications that enable the automated tracking of the mobility patterns of their users. Apart from supporting fully automated tracking, the applications also enable users to manually provide additional feedback on their trips. To systematically evaluate the collected data, we have developed several data analysis tools. In this paper, we describe the mobile applications as well as the analysis tools from a technical perspective. To preserve the privacy of the users, the mobile applications implement an on-device segmentation that identifies mobile and stationary segments to obfuscate the latter before uploading the trips to a web-service. Using the applications, we have collected more than 13,000 trips from 161 users over a period of 3.5 months. Since the participants of the field trials have manually provided information on approximately 4000 trips, we rely on machine learning to automatically derive the information for the remaining 9000 trips. To do this, we train a hierarchical classifier consisting of two random forests. Our evaluation suggests a resulting accuracy of 89.7% for identifying the transport mode of short trip segments and 95.6% for complete trips while differentiating walking, cycling, driving, public transportation and e-scooter rides.