Research And Application Of Campus Card Consumption Based On Data Mining

In recent years, campus informationization has become a mainstream trend. Especially in Colleges and Universities, the campus card is the main carrier of the campus information construction and the campus card can achieve a variety of functions, therefore, it is known as the "universal card". Every day will produce a large number of consumer data, such as Internet access, dining, shopping. There are a lot of valuable consumer rules or information hiding in these chaotic data and these information have important meanings to the management of colleges and universities. Therefore, how to dig out the valuable information has become a hot topic of today's attention. The development of data mining technology provides technical support for solving this problem.Based on the study of Mining Data and WEKA platform, by using consumption data of the undergraduates of Lanzhou Jiaotong University. This article will dig out students' consumer behavior characteristics by clustering and associating. And related to the poor student library data for the students to make assessment of the issue of poverty. So as to provide reference for the relevant departments of the school management decision-making. This paper mainly research and analysis from the following aspects:(1) Aiming at the disadvantages of traditional K-means algorithm, such as the K value and the initial center point are highly dependent, this paper proposes an improved VW-K-means clustering algorithm. In the initial stage to join the voronoi chart, adaptively, resulting in a better K value. Secondly, in order to reduce the sensitivity of the algorithm to noise data and avoid the data objects of the low density and large cluster edges are error divided into high density small clusters. In the weighted average method based on the criterion function to join the weight coefficient value to effectively divide the data objects, reduce the probability of false division. According to the clustering of UCI data set, the experimental results show that the improved VW-K-means algorithm has higher clustering accuracy, faster convergence, and achieve the global optimum, which improves the performance of the traditional algorithm.(2) In this paper, we select the data of campus card from July 2014 to March 2014. At first, a lot of work is to pre-process the data, and then based on the data mining algorithm to make use of the data analysis.First, what must to do is statistical analysis of the number of dining hall, student dining options and student consumption behavior. And the results of statistical analysis and cluster analysis are compared to facilitate more accurate research and analysis.Secondly, based on WEKA platform and the VW-K-means algorithm from two aspects to make mining analysis: On the one hand, making the cluster analysis of the number of diners in the school canteen. Thus the school cafeteria can be based on this data to carry out a reasonable planning, so as try to avoid the crowd cafeteria and the waste of food. On the other hand, through the analysis of the students' consumption level, the consumption range can be divided reasonably, which can be the foundation of the research of association rules.Finally, based on the WEKA platform and Apriori algorithm, to find out the hidden connection between the consumption situation of the students and the actual situation of the poor students by associating them. Using J48 classifier based on WEKA platform, the students' consumption level is divided into three categories to predict the students' poverty subsidy and for our school related departments to provide a reference for the decision of the data base.