Research And Application Of Education Data Mining Based On Decision Tree Technology

With the enforcement of the nine-year compulsory education system,some relatively developed regions have brought high school education into the scope of compulsory education,thus carry out twelve years of compulsory education.The enrollment expansion policy of colleges and universities,makes more students have an opportunity to study in university,so high school education has become a necessary stage for most of students.This has led to the accumulation of a large amount of student data in the educational administration system.And the data is exponentially growing and takes up a lot of storage space.If we can take advantage of these data,we can not only better learn the situation of students,but also predict the future partially and make auxiliary decisions.For example,according to students' data,teachers can not only learn the current situation of students,but also teach students according to their aptitude based on the result of analysis,so as to improve the teaching effect.As an important branch of data mining,education data mining(EDM)is about how to dig out potential and valuable information from a massive student data,has attracted relevant scholars' attention.Education data mining is extract valuable information from the massive educational data by using mathematical methods and computer technologies to help us raise teaching quality and educational and management ability.The mainly work of this thesis is to use the decision tree technology to analyze the multi-label decision tables and optimize the storage of the resulting decision trees.According to the common multi-label decision table in the educational administration management system,this thesis analyzes the factors that affect the students' achievement and predicts the students' scores based on the decision tree generated.Furthermore,measures should be taken to improve students' learning efficiency,teachers' teaching skills.Learning the situation of the students by the prediction results is of great significance improve the educational work,to promote the teaching effect and to cultivate talents.This thesis first discusses the research background and significance,and introduces the research status and development of educational data mining and decision tree technology at home and abroad.And it enumerates the extensive application of educational data mining in real life.Secondly,this thesis gives the concept of educational data mining and describes several common decision tree algorithms(such as ID3/C4.5/CART/SLIQ)and their relations and differences.Then,the concepts of multi-label decision table and decision tree are given in detail,and the decision tree analysis method of multi-label decision table is proposed.To a large extent,this method takes dynamic programming as the core idea,and proposes an algorithm is proposed that minimizes the size of decision trees,and then by which the valuable information is extracted from multi-label decision tables.Finally,in view of the size of the generated decision tree,therefore,a recursive algorithm is proposed to combine identical subtrees and identical leaf nodes to decision diagram.The obtained decision diagram has no redundant node is small in size,its storage space is reduced.The main factors influencing the students' achievement are found out in the decision diagram,which provides a reference for improving the school enrollment rate.