Research And Analysis Of Student Behavior Profile Based On Big Dada

In information management system on campus,with large scale behavior data of students increasingly detailed and complicated,analyzing and mining these various behavior data of students motivate a novel innovation in the current educational fields,at the exactly same time,constructing student behavior profile finds the relationship between behavior data of students and real life.Student behavior profile refers to making the use of the semantically oriented label sets of behavior data to identify behavioral characteristics of students and describe behavior habits of students.This paper proposes a research and analysis of student behavior profile based on big data.The study is mainly based on various behavior data of students,spatial clustering algorithm model and classification algorithm model based on association rules as a supplement,and is mining potential value hidden behind student behavior.So then,in order to meet the requirement of behavior profile of students and break the preconceptions,the paper aims at building guidance and forecasting for behavior and orbit of students and helps to improve the conventional teaching and learning pattern.The research contents of this study mainly include the following aspects:(1)Process behavior data of students.Owing to that the problem that the storage and magnitude of data of students are different due to various sessions of behavior and orbits,a strategy preprocessed and divided behavior data of students is established.Therefore,the relatively complete data of students are used for later research based on integration of dynamic behavior and static attributes,such as gender and major,Score,Monetary,Frequency and Borrowing.This process eliminates the differences of different orders of magnitude of various behavior data of students,and as such normalizes and integrates various behavior data of students.(2)Analyze various behavior data of students.Based on the normalized and integrated various behavior data of students with united orders of magnitude,the paper combines the improved Elbow method and K-means clustering algorithm to cluster various behavior data of students.First,the algorithm is respectively applied on various behavior data of students,namely,Score,Monetary,Frequency and Borrowing,and the optimal k-value and clusters of Score,Monetary,Frequency and Borrowing are respectively determined.Then this paper combines the anti-normalization values of various behavior data of students with the clusters of Score,Monetary and Frequency and Borrowing,which is used for digitized description of various behavior data of students.Finally,in terms of the digitized description and different clusters of various behavior data of students,the discretized label sets of various behavior data of students are further constructed.(3)Construct behavior profile of students.Based on different clusters of various behavior data of students,classification algorithm model based on multiple frequent pattern tree is proposed by integrating static attributes of students and the discretized label sets of various behavior data of students.And the frequent patterns and the tree structure of discretized labels sets of various data of students are found and determined,which identifies and describes the prefix path for various behavior data of students to construct student behavior profile.Experimental results show that student behavior profile constructed in this paper can effectively guide and forecast the developing tendency of data for students.