Analysis Of Users' Personnel Correlation With Trajectory Based On Data Mining

In recent years,the application of Internet and LBS technology have been developing rapidly.And the informationization of higher education has been pushing forward constantly.What has been known to us is that more and more universities are now beginning to use Wi-Fi for teaching management.And for a growing number of Wi-Fi users' data in campus,it will do much help if we use these data properly.Under this background,we have analyzed the behaviors of Wi-Fi users in campus in this thesis,including studying the group patterns of users and constructing a model for data mining.Our goal is to analyze the correlation of Wi-Fi users in campus,to study their characteristics of trajectory and to do clustering work about them.This thesis firstly analyzes and presents the pattern of users' location distribution,time distribution and some other characteristics.Based on the statistical information above,we then collect data and group data by time and location to get the dataset for correlation work.We then transform the origin data and classify it by the id of different locations.After the work of feature extraction,we finally get the matrix of users' distribution and we get Wi-Fi users' traveling trajectory.For the dataset of mac addresses,we use a fast algorithm which is based on frequent pattern tree to analyze the correlation of different mac addresses.After the correlation analysis,we calculate the trajectory similarity between different users.Meanwhile we have achieved the recovery and comparison of users' trajectory.At last,we use the hierarchical clustering algorithm which generates from bottom to top to achieve the clustering work of Wi-Fi users in campus,and finally we have got 13 types of personnel in the campus we study.