Understanding campus mobility

• Main Author: Toh, Christopher Gerard Wei Hong
• Other Authors: Lee Bu Sung, Francis
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Computer science and engineering::Data; Social sciences::General::Education
• Online Access: https://hdl.handle.net/10356/137905
• Institution: Nanyang Technological University

As tertiary education institutes improve their quality of education, they seek advancement in technology. Thus, people on campus are constantly connected to the network, resulting in collections of large amounts of network connection data which can be used to understand human mobility on campus. This project sets out to accomplish two things: provide greater understanding on the possibilities that such network data can be used for through data analysis and visualisation; and study behavioural and social patterns amongst student athletes through group identification. This is achieved by rigorous pre-processing of raw data attributes obtained from Cisco Wi-Fi access points collected by Nanyang Technological University (NTU) Centre of Information Technology Services around campus. Data preparation, analysis, and visualisation scripts are written in Python. To begin with, an exploratory study on mobility patterns on campus, looking at spatial and temporal aspects of mobility. A main finding would be that the arrival rate of people on campus increases at a faster rate, with counts peaking between 1100H and 1300H, before decreasing at a gradual rate thereafter. This project also finds that it is possible to build on the current pre-processing steps by conducting user-device profiling, mapping activities to locations, and point-of-interest location analysis. To understand social relations better, identifying group behaviour amongst students is necessary, and student who partake in sports-related activities (s_students) were chosen as a pilot study. The project finds that s_students spend a large amount of time amongst other s_students, totalling to 32.9% of their time on campus, which is more than a general student (g_student) who spends 8.4% of time in social groups. This indicates that s_students are more likely to form cliques and groups amongst themselves and suggests the possibility that the sports activity provides a strong platform for friendship bonds to be formed. With these findings, this project looks to propose several next steps – such as using this analysis to make better management decisions, and using unsupervised learning techniques for group identification – which could take the analysis of human mobility to the next level, building on the work that has already been done.