| Nowadays there is a great hidden danger in college students’ physical and mental health.It is necessary to find out college students’ health problems in a timely manner.Smart campus construction produces a large amount of data,these data contain a lot of knowledge of students,which implied the health status of college students.In order to make full use of data created by smart campus,and find out college students’hidden sub-health problems.This article puts out a research on college students’sub-health mining method based on the smart campus environment.According to analyzing students network behavior,building students portraits,to analysis students of sub-health state objectively,based on the following respects:(1)By using the IC card,access control,network log data,extracting students’activities trajectory vector set in the school and achieve a structured students’trajectory data.Different life represents different living habits and it can give a reaction of sub-health state in a certain degree.Using the improved trajectory frequent pattern mining algorithms to dig out activities frequently path.Finally we can achieve a normalization life trajectory→the health level according to the result set.(2)By using the network log data to extract students online use vector set.By analyzing the online time,online length,up and down the line frequency,stay up late to find the network narcissism degree.Finally,we can achieve a normalization network using habits→the health level according to the result set.(3)By using the consumption data to extract students consumer behavior vector set.By analyzing the consumption amount,the number of key indicators data and so on to find out the irregular diet information.Finally,we can achieve a normalization economic level→the health level according to the result set.(4)By extracting course performance,course study and book borrowing data,we select some key attributes and construct a multiple linear regression model.Then using the model to predict students learning stress and risks of hanging.Finally,we can achieve a normalization study level→the health level according to the result set.(5)By extracting the spatio-temporal consumption data,and using an improved frequent pattern mining method to analysis the relationship between college students to find out the interpersonal relationship data.Finally,we can achieve a normalization relationship level→the health level according to the result set.In the end,we build a student portrait based on the above five aspects comprehensively.As a result,we can illustrate the sub-health level of students from a comprehensive point of view to explore the wisdom of the campus data. |
