基于学生在线学习行为特征的混合课程分类研究

Research on Blended Courses Classification Based on Characteristics of Students Online Learning Behavior

课程分类对其设计、实施及评价十分重要。混合课程的动态设计和实施、个性化评价以及学习预警研究都要求数据驱动的课程分类,然而这种分类方法目前尚在探索中。该研究选取某高校网络教学平台中2018年秋季学期2456门混合课程的在线数据作为样本,提出了一种依据学生在线学习行为聚类特征对混合课程进行分类的方法,并采用2020年春季学期的1851门混合课程对该分类方法的稳定性进行了验证。结果表明:(1)该方法通过机器学习算法对混合课程中的学生在线学习行为进行聚类并提取每类学生的典型特征,并据此将混合课程分为可以自动识别的五种类型:不活跃型课程、低活跃型课程、任务型课程、阅览型课程和高活跃型课程;(2)采用该方法对同一个高校两个学期的混合课程进行分类,结果都归入了五个类别之中,且每类课程中学生学习行为的典型特征相同,由此验证了该方法具有良好的稳定性;(3)该方法不依赖人工事先标注,便于计算机自动化分类,能发现课程中的学生群体行为特征,分析学习过程差异,为教师动态设计、实施混合课程,及时预警学生并实现个性化混合课程评价奠定基础。

Classification of blended courses is crucial to the design,implementation and evaluation of the courses.At present,research on dynamic design and management,individualized assessment,and early intervention in blended learning all requires datadriven classification of the blended courses.Yet,the classification methods are still under exploration.This study extracted the data of 2456 blended courses from the learning management system of a university in fall 2018 semester,proposed a classification method of blended courses based on clustering characteristics of student online learning behaviors using this data sample,and tested the stability of the method with data of 1851 blended courses in spring 2020 semester.The result shows that(1)This method performed cluster analysis on student online learning behaviors in blended courses with machine learning algorithms,and recognized the typical pattern of students in each cluster.Accordingly,blended courses can be classified into five auto-identifiable categories:Inactive Courses,Low Active Courses,Task-Related Courses,Reading-Related Courses,and High Active Course.(2)Using this method,all blended courses of a same university across two semesters fall into one of the five categories and typical patterns of student learning behavior of the courses in each category remain the same,which proved the stability of the method.(3)This classification method does not require manual data labeling,thereby realize automatic classification.The method identifies the student group behavioral patterns in blended courses,analyzes the differences in the learning process,and provides reference for dynamic design and management,early intervention,and individualized learning in blended courses.