Exploration of University Students’ Mobility Behavior on Campus

Abstract

It is a subject worthy of being studied to predict human mobility through the big data of human movement trajectory. The prediction has been widely used in many fields. In this subject, we record and analyze the movement trajectory of students on campus during their university years with a purpose to make a prediction on mobility based on Markov chain models. We explore from the Campus Smart Card of the university, which records most of the activities of a university students on campus, and locks its position. Specifically, our data set include about 16.8 million consuming logs came from 4,741 students, they study at school from Sept, 2015 to June, 2019. The predictability differences among different students are smaller than those of individuals in Call Detail Records (CDR) data, which means that students seem more predictable. We also make predictions based on Markov chains model of different orders and find that high order Markov chains have better performance, although the inherent reason for this needs further research. Our work provides sufficient support in predicting human mobility area.