Social Network Analysis Based on Canteen Transaction Records

College students' social network is critical because it could influence students' academic performance, attrition, and even suicide attempt. Unfortunately, the college students' social network is not easy to be collected. The existing methods (e.g., survey, online social network web sites, and phone call records) suffer from various shortcomings. In this paper, we present a case study, in which the college students' social network is extracted from their canteen transaction records. To this end, we empirically collect a canteen transaction data set which includes 4.5 million transaction records of 16 thousand undergraduate students during a semester (112 days). We propose a systematic method to extract the social network. Based on the extracted network, we calculate some network attributes for each student, and employ the regression analysis to study the relationships between the network attributes and students' academic performance. The findings in this case study encourage us to build more rigorous statistical methods to extract networks from transaction records, and to examine the effect of network attributes