Analysis And Prediction Of Student Behaviors Based On WiFi Check-in Data

With the coming of the era of big data,more and more research has been done on college students.And the rapid popularization of smart mobile terminals in university campuses has made universities accumulate a large amount of Location-based Social Network(LBSNs)data,which provides an objective condition for students' behavior analysis.In this paper,two kinds of Campus Geographic Information based Logistic Matrix Factorization(CGLMF)and Dense Data Matrix Decomposition Embedding(MFED)are proposed using students WiFi check-in dataset for Point-of-interest(POI)recommendation.And these algorithms experimented on students WiFi check-in dataset and visualized in the campus WiFi check-in data analysis system.Specifically,the main contributions as follows:(1)Based on the existing campus geographic model and POI recommendation algorithm which fuses geographic information,this paper presents Campus Geographic Information based Logistic Matrix Factorization(CGLMF)POI recommendation algorithm.This algorithm utilizes students' personal information and campus geographic information,considers the main activity areas of students and the correlation of each POI in this area,and then incorporate the geographic information model into Logistic Matrix Factorization to improve the POI recommendation performance.(2)Based on the existing POI neural network personalized embedding algorithm and students WiFi check-in dataset characteristics,a Matrix Factorization Embedding for Dense Data(MFED)algorithm is proposed,which embeds the relationship between students' personal preference and the correlation of each POI in the dataset into the neural network to implement more accurate personalized POI recommendation.(3)Visual analysis using the campus WiFi data check-in analysis system design by campus GIS map system.Starting from design overview,system requirements analysis,to system function implementation.The system designs different charts for user mobility analysis and modeling in general user statistics and behavior data to show the data characteristics and the result of POI recommendation.The system focuses on the display of the effect of CGLMF POI recommendation algorithm and MFED POI recommendation algorithm.