决策树技术在高校学生成绩分析中的应用分析-application and analysis of decision tree technology in college students performance analysis.docxVIP

摘 要由于高校的连年扩招，各个学校都积累了大量的学生基本数据信息和学习成绩数据 信息，但是这些数据都还停留在备份、查询、简单的统计的阶段，不能发挥其应有的作 用。这样就导致大量的成绩数据只是以各不相同的表格形式存储于不同的计算机上，不 利于对这些数据所隐含的信息进行充分的挖掘分析，不便于教学管理部门进行有效的教 学评价，也不利于教师对其所带学生的学习情况进行全面的指导。本文主要应用数据挖掘技术中的决策树技术对高校学生的成绩进行分析，找出影响 学生成绩取得“优秀”和“差”的因素。在建立分类模型时，以信息增益率作为属性选 择的度量标准，对学生成绩是否为“优”的数据训练集应用决策树技术中的 C4.5 算法 来进行模型的建立，从而生成分类规则，得出影响学生成绩为“优”的因素。C4.5 算法 中以信息增益率作为属性选择的度量标准时，在计算属性的信息量时涉及到了对数运 算，计算时间效率低下。对计算属性的信息量公式进行改进，形成新的信息量计算方法。 应用改进后的 C4.5 算法对属性进行信息增益率的计算，并且对学生成绩是否为“差” 的数据训练集进行模型的建立，生成分类规则，得出影响学生成绩为“差”的因素，为 教师有针对性的提高教学质量提供了有力的数据支持。关键字：数据挖掘决策树技术C4.5成绩分析AbstractDue to successive years of enrollment of the university, every school has accumulated a large number of students’ basic data and achievement data, but these data are still stuck in the backup, the query, the stage of simple statistics, and can not play its due role.This led to a lot of performance data is only stored on different computers vary in tabular form, and those data can not conducive to the information implied by these data mining analysis,can not to facilitate the teaching departments for effective teaching evaluation, and can not conducive to comprehensive guidance teachers brought student learning.In this thesis,through the analyze of the college students’ achievements based on decision tree technique,we get the factors of identify the achievements of college students affect excellent and bad .in creating a classification model, we choose the the rate of information gain as attribute selection metrics.we use C4.5 algorithm to set to the model for the scores are “excellent”,and generate the classification rules and obtain the factors of the student performance as excellent . C4.5 algorithm used the gainrate of information as the attribute selection metrics,and related to the logarithm in calculating the amount of information ,then it is inefficient in time.Here,I improve the formula for calculation of property the amount of information and form a new method of calculate information. In creating a cla