基于大学课堂学习行为的成绩预测方法研究

Research on the Performance Prediction methods based on Classroom Learning Behavior

文章首先提出了一种基于大学课堂学习行为的成绩预测方法。随后,文章以武汉市H高校2016~2019年11门本科课程的相关数据和考试成绩为例,通过最终考试成绩与相关行为变量的相关性分析,确定预测变量,并设计了4种预测变量组合。最后,文章采用常用的四种机器学习分类算法,对这4种预测变量组合的精度进行了对比分析,结果表明:使用LMS行为变量和课堂视频行为变量组合的预测准确率最高,但仅有理论价值;而LMS行为变量和计算机自动分析课堂视频行为变量的组合方案虽然准确率不算最优,但优于单纯使用LMS行为变量进行预测的准确率,是一个精度较优、实施性强的推荐方案。

Firstly,this paper proposed a performance prediction method based on college classroom learning behavior.Then,taking relevant data and final exam scores of 11 undergraduate courses from 2016 to 2019 in H University in Wuhan for example,this paper determined the predictive variables and designed four combinations of predictive variables through the correlation analysis of final exam scores and related behavioral variables.Finally,using four commonly used machine learning classification algorithms,this paper compared and analyzed the accuracy of four combinations of predictive variables.The results showed that using the combination of learning management system(LMS)behavioral variables and classroom video behavioral variables had the highest prediction accuracy,but which only had the theoretical value.Although the accuracy of the combination scheme of LMS behavioral variables and computer automatic analysis of classroom video behavioral variables was not optimal,but which was superior to the accuracy of using LMS behavioral variables alone,and it is a recommended scheme with better accuracy and strong implementation.