大学教学评价数据的挖掘方法研究

发布时间：2018-01-19 15:41

本文关键词： 教育数据挖掘教学评价离群点检测关联规则　出处：《河北师范大学》2017年硕士论文　论文类型：学位论文

【摘要】：“数据驱动学校,分析变革教育”的大数据时代已经来临,利用教育数据挖掘技术和学习分析技术,构建教育领域相关模型,探索教育变量之间的相关关系,为教育教学决策提供有效支持,将成为未来教育的发展趋势。课堂教学质量评价是教学质量评估的重要内容,教学评价是教学活动中不可缺少的环节。现有的教学评价结果大多采用一种定量评价方法,该方法通过量化计算分析,部分的反映了教学状况,但很难直接看出这些数据之间是否还隐含着更有价值的规律,也就很难为教学管理者的决策提供更有效的支持。本文结合教学评价数据,研究相应的知识发现方法,从教学评价数据中挖掘隐藏的知识,给教学管理工作者管理教学和教师授课提供意见。主要的工作内容有:(1)教学评价数据的数据分析和预处理。完成了教学评价数据的数据梳理工作,分析了教学评价数据的数据结构,构建了教学评价数据相关表的关系结构图;完成了数据的的初步统计分析和数据预处理工作,完成了对教学相关表的数据统计工作,根据确定的挖掘目标选取实验数据并完成了数据的清洗、集成、变换和规范化,通过对数据进行预处理,将数据处理成挖掘所需的数据格式。(2)教学评价数据的离群点检测算法研究。本文分析了现有的分类型属性离群点检测算法的优缺点,及其应用于教学评价数据上的不适应性,提出了一种教学评价数据的离群点检测算法,并将此算法应用于真实教学评价数据,检测教学评价数据中的噪声数据将其分离至单独的数据表中。实验表明算法具有较高的效率和准确率。(3)教学评价数据的关联规则挖掘算法研究。研究了一种高效的教学评价数据关联规则挖掘算法,以发现教学因素间有哪些联系,从而给教师的教学,和教学管理者的教学管理工作提供建议。本文分析了经典Apriori算法的不足,并分析了现有的基于压缩矩阵的关联规则挖掘算法用于多值分类型数据时的不适用性,提出了一个新的多值分类型属性关联规则挖掘算法。实验表明,算法可以有效的避免无效候选项集的产生,减少了不必要的连接操作,提高了挖掘效率。
[Abstract]:The big data era of "data-driven school, analysis of education for change" has come. Using the technology of educational data mining and learning analysis, the relevant models in the field of education are constructed, and the correlation between educational variables is explored. Providing effective support for educational decision-making will become the development trend of education in the future. Classroom teaching quality evaluation is an important content of teaching quality evaluation. Teaching evaluation is an indispensable link in teaching activities. Most of the existing teaching evaluation results adopt a quantitative evaluation method, which partly reflects the teaching situation through quantitative calculation and analysis. However, it is difficult to directly see whether there is a more valuable rule between these data, and it is difficult to provide more effective support for the decision-making of teaching managers. The corresponding knowledge discovery method is studied, and the hidden knowledge is mined from the teaching evaluation data. The main contents of this paper are: 1) data analysis and preprocessing of teaching evaluation data. The data combing of teaching evaluation data has been completed. The data structure of teaching evaluation data is analyzed, and the relation structure diagram of the related table of teaching evaluation data is constructed. Completed the preliminary statistical analysis of the data and data preprocessing, completed the data statistics of the teaching related tables, selected the experimental data according to the determined mining targets, and completed the data cleaning and integration. Transform and normalize, by preprocessing the data. This paper analyzes the advantages and disadvantages of the existing outlier detection algorithms based on the analysis of the characteristics of the existing outlier detection algorithms. This paper presents an outlier detection algorithm for teaching evaluation data, and applies the algorithm to the real teaching evaluation data. The noise data in the teaching evaluation data is separated into a separate data table. The experiment shows that the algorithm has high efficiency and accuracy. The research of association rules mining algorithm for teaching evaluation data. An efficient association rule mining algorithm for teaching evaluation data is studied. In order to find out the relationship between teaching factors and provide suggestions for teachers' teaching and teaching management, this paper analyzes the shortcomings of classical Apriori algorithm. The inapplicability of the existing association rules mining algorithm based on compressed matrix for multi-valued classification data is analyzed, and a new algorithm for mining association rules of multi-valued sub-type attributes is proposed. The algorithm can effectively avoid the generation of invalid candidate itemsets, reduce unnecessary join operations, and improve the efficiency of mining.
【学位授予单位】：河北师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：G434;TP311.13

【相似文献】