基于核PCA的混合教学数据特征分析与分级

Feature Analysis and Classification for Mixed Teaching Data Based on Kernel PCA

针对目前教学数据挖掘与分析多考虑在线学习特征数据,而缺乏对传统课堂和在线教学的混合数据的融合分析,所建立评价模型也多以教师为主体,鲜少考虑学生个体质量分级等问题,本文研究了混合教学数据的非线性群集特征提取方法,采用核主成分分析法对非线性混合学习特征数据进行降维计算,根据特征的权值大小提取影响学生学习质量的主要特征群集并得到多级特征指标,构建混合教学模式下学生个体质量的综合评价指标体系,并采用综合评价法确定多级评价指标的权重,建立学生个体综合评价与分级模型,并以本校公共课的混合教学数据为例,验证了模型的有效性。

At present teaching data mining and analysis mainly consider online learning characteristic data,but lack of fusion analysis of mixed data including traditional classroom and online teaching.In addition,the evaluation models established are mostly teachercentered and seldom consider the individual quality classification of students.Therefore,the method of nonlinear cluster feature extraction for mixed teaching data is studied,and the dimensionality reduction of nonlinear mixed learning characteristic data was calculated by using kernel PCA to construct the comprehensive evaluation index system of students'individual quality with the mixed teaching mode.Meanwhile,the weight of multi-level evaluation index is determined by the comprehensive evaluation method,and then the individual comprehensive evaluation and classification model of students is established.Finally,taking the mixed teaching data of public courses in our university as an example,the validity of the model is verified.