一种基于C4.5决策树的贵州省高校贫困生评定方法  

Assessment of Poor College Student in Guizhou Province via C4.5 Decision Tree

作　　者：李明江[1] 卢玉[1] 刘彦[1] 

机构地区：[1]黔南民族师范学院计算机科学系,贵州都匀558000

出　　处：《科技通报》2013年第8期223-224,233,共3页Bulletin of Science and Technology

基　　金：贵州省教育厅自然科学研究青年项目(黔教科20100095)

摘　　要：提出了一种基于C4.5决策树的贵州省高校贫困生评定方法。首先从贵州省大学生的消费行为、家庭情况、贷款与助学行为3个方面建立了大学生贫困资格评定的指标体系;其次,将获得的15项指标作为C4.5决策树的特征属性,基于信息增益率完成对连续变量的离散化处理,将知识表示成树的形式,采用错误预测率进行修剪,得到了影响贫困学生评定的4个最重要变量;最后将该方法进行实证分析。结果显示,它不仅原理简单,解释直观,而且计算快速准确。相比同类方法,它不依赖于数据的统计分布,也不需要选择模型参数,是一种有效的高校贫困生分类评定技术。A C4.5 decision tree based assessment approach of poor college students in Guizhou province was proposed in this paper. Firstly the index system was established from the consumer behavior of students, the economic condition of their families and their work-study status. Secondly, 15 indexes are taken as the attributes of data to be classified by C4.5 decision tree, and the continuous attributes are discrete according to the information gain-ratio of attributes. The tree is pruned using the prediction error to obtain the four most important attributes to characterize the poor students. Fi- nally some real data is used to validate the efficency of our proposed method, and the experiments results show that it is of simple principle, and cgariacteristic of rapid and accurate calaulation. Compare with its counterparts, it not only does not rely on the statitieal distribution of data, but also need not choose the model parameters, so it is an efficient technolo- gy for assement of poor college students.

关 键 词：贫困生评定 指标体系 决策树 C4 5 

分 类 号：G526.78[文化科学—教育学]