Research And Application Of Student Behavior Analysis Based On Consumption Data Of Campus Card

In order to better mine the hidden knowledge hidden in the students' one card consumption behavior and assist the school,teachers and students' managers to make more accurate decisions,this thesis designs and implements the one card consumption behavior analysis system.In order to make the system more efficient and accurate,two data analysis methods,mathematical statistics and data mining,are used to mine and analyze the data of one card consumption,to explore the law of students' one card consumption behavior and the potential relationship between consumption behavior and life,learning,employment and safety.The details are as follows:1.PVW-kmeans algorithm is designed to be used in the cluster analysis engine module of the system.Based on the K-means algorithm,the algorithm has made three improvements and achieved better clustering effect.Specifically,it includes:(1)based on the improvement of Tyson polygon,it can more accurately select the initial clustering center point and determine the number of initial clusters;(2)based on the improvement of weighted average,it can more accurately distinguish each data object and cluster data objects faster;(3)based on the improvement of principal component analysis,it can reduce the dimension of multi-dimensional consumption behavior characteristics of all in one card.And use PVW-kmeans algorithm to cluster analysis of students' consumption behavior,and make an experimental comparison between the improved PVW-kmeans algorithm and K-means and K-means + + algorithm to verify the efficiency and accuracy of PVW-kmeans algorithm.2.WMDE-Apriori algorithm is designed for association analysis engine module of the system.Based on Apriori algorithm,the algorithm is improved in three aspects to obtain more effective association rules.Specifically,it includes:(1)based on the improvement of weight and matrix,it reduces the number of candidate set and repeated scan of transaction database;(2)based on the improvement of data division,it eases the bottleneck of algorithm operation pressure,and improves the efficiency of algorithm;(3)based on the improvement of influence degree,it makes the association rules more strengthened and effective.The WMDE-Apriori algorithm is used to analyze the relationship between students' consumption behavior,and the improved WMDE-Apriori algorithm is compared with Apriori algorithm in the experiment to assist poor studentsto identify,which verifies the effectiveness and efficiency of WMDE-Apriori algorithm.3.This thesis designs and implements a consumption behavior analysis system for all-in-one card,which includes:(1)overall design of the system architecture,system functions,database,data flow,system interface,etc.;(2)detailed design of the data sorting module,statistical analysis engine module,clustering analysis engine module,association analysis engine module,internal interface of the system,auxiliary recognition module for poor students and statistical analysis module.The system shows the rules of students' consumption behavior and the implicit relationship with other behaviors by visual technology,and gives an alarm to abnormal behaviors in time,which not only facilitates the management of the school,teachers and students' management staff,but also provides strong data support to assist their decision-making.