摘要
在课程学习中实时对学生掌握情况进行预测,可以尽早对学生学习行为进行干预。本文通过在线互评系统采集学生知识学习中与最终考试成绩相关的课程参与度、作业提交频次、作业完成时间、自评分数以及匿名互评分数等相关性预测变量,预测变量的采集全程无监督、学生无感知、平时成绩无教师校准评价干预,提出一种基于在线互评的成绩预测模型研究。对比常用的机器学习分类算法,通过实验分析可知,多元线性回归算法在预测变量不确定的前提下,在算法稳定性、准确性上都能满足可靠预测需求。
In the course of study, real-time prediction of students’ mastery can intervene in students’ learning behavior as early as possible. This paper uses an online mutual evaluation system to collect relevant predictive variables such as course participation,homework submission frequency, homework completion time, number of self-ratings, and anonymous mutual ratings related to the final exam scores in student knowledge learning. The collection of predictive variables is unsupervised throughout, Students have no perception, and there is no teacher calibration and evaluation intervention in normal performance. A performance prediction model based on online mutual evaluation is proposed. Compared with commonly used machine learning classification algorithms,experimental analysis shows that the multiple linear regression algorithm can meet the demand for reliable prediction in terms of algorithm stability and accuracy under the premise of uncertain predictor variables.
作者
魏明俊
WEI Mingjun(Hubei University of Medicine,Shiyan Hubei 442000,China)
出处
《信息与电脑》
2021年第23期74-76,共3页
Information & Computer
基金
湖北医药学院“基于数据挖掘的学习预警系统的设计与实现”(项目编号:2020QDJRW011)。
关键词
在线互评
相关性分析
学习行为
成绩预测
多元回归
online mutual evaluation
correlation analysis
learning behavior
performance prediction
multiple regression
