The Application Of Decision Tree And Association-rule Algorithm In Colleges’ Endorsement And Management Of Impoverished Students

With the continuous development of education undertakings, the enrollment ofcolleges and universities has expanded rapidly. Thus, the pressure of poor studentsincreases obviously. The evaluation criteria and method of endorsement is the key tothe work of endorsement. At present, most colleges adopt artificial methods, whichare less efficient and exist many disadvantages, such as nontransparent andnonobjective evaluation criteria. Therefore, the research on endorsing method forpoor students in colleges has certain practical value. Based on data mining technology,the research on using the algorithms of association rules and decision tree to extractvaluable information from huge amounts of data is of great significance.Combining association rules with decision tree algorithms, this paper proposesa new decision tree method based on association rules and then applies this method toendorse poor students. Firstly, we analyze the implement steps of the two algorithms.Since Apriori algorithm has high performance and highly reduces the size ofcandidate sets, and C4.5algorithm has high classification precision among decisiontree algorithms, the proposed algorithm is actually C4.5algorithm based on Apriorialgorithm. Secondly, after describing the process of constructing a decision tree, weintroduce the association rules algorithm into the process of constructing the decisiontree. Using a series of rules generated by association rules algorithm to construct newattributes, we can reorganize the dataset. Thirdly, the C4.5algorithm is improved.Trough merging those branches with higher information entropy value into thebranches with lower value, it effectively avoids fragmentation issues of C4.5algorithm. In addition, we introduce a balance coefficient called ω in the calculatingprocess of information entropy. By specifying a balance coefficient for some lessimportant attributes, their information entropy can be relatively reduced. Thus it cangenerate a decision tree with a higher accuracy. At last, the improved algorithm isused to classify a testing set. To illustrate advantages of the improved method, threedatasets from UCI database are used in experiment. The results show that theimproved method has obvious advantages compared with the original algorithm.This paper is aimed at constructing a decision tree model for endorsingimpoverished students and then evaluates the model.655data records from3majorsare selected to evaluate the model. After comparing with the existing results, theexperiment results show that the consistent proportion of the two results is as high as87.48%. High efficiency shows that the identified model is of great significance to the future actual work of endorsing impoverished students.