学习绩效预测模型的应用:运动与课堂行为关系研究被引量：9

Using learning performance prediction model to predict the relationship between physical exercise and classroom behavior

原文传递

导出

摘要学习行为诊断、监控与评价一直是教育数据挖掘的研究重点。学习绩效预测研究已将数据模型的理论转化为教学的应用实践,而融合多种行为数据对学习绩效进行预警与评价的研究较少。采集1,053名大学生12周的运动日志和课堂视频,构建运动与课堂行为预测指标,利用决策树和规则集方法构建预警阈值区间和行为预警策略,并利用机器深度神经网络分类模型,结合主成分法和熵值法,得到两种行为的权重和得分区间,构建行为组合评价策略。研究表明:学习绩效模型能有效预测运动与课堂行为对学习绩效的影响,学习行为预警策略也能有效发现学生行为变化规律,学习行为组合评价策略可量化行为特征值与学习绩效的关系,提升教学管理能力与教育治理水平。 Learning behavior diagnosis,monitoring and evaluation has attracted significant research attention from the field of educational data mining.Research on learning performance prediction has shifted from data modeling to application in reality.Nevertheless,there is little research integrating multiple types of learning behavior data to predict and evaluate learning performance.This study collected the exercise logs and classroom learning videos of 1,053 students in a 12-week period.It developed predictive indicators of physical exercise and classroom behavior,adopted decision trees and rule algorithms to generate intuitive and readable decision tree graphics and rule sets,and constructed early-warning threshold intervals and behavioral early-warning strategies.The machine deep neural network classification model was used together with the principal component and entropy method to obtain the weight and score interval of the two behaviors and develop the strategy of behavior combination evaluation.Findings show that the learning performance prediction model can effectively predict the impact of physical exercise and classroom behavior on learning performance,that early-warning strategy for learning behavior can effectively identify patterns of change in leaning behavior,and that the strategy of behavior combination evaluation can quantify the relationship between behavior characteristic values and learning performance,hence able to improve teaching management and education governance.

作者胡航李雅馨 Hang Hu;Yaxin Li

机构地区西南大学教师教育学院卓越教学中心西南大学学习教学与智能化研究中心西南大学教育学部

出处《中国远程教育》 CSSCI 2021年第10期19-28,76,共11页 Chinese Journal of Distance Education

基金中央高校基本科研业务费专项资金重点项目“基于多模态数据源的深度学习数据库建设及发生机制研究”(项目编号:SWU2109236)。

关键词学习绩效预测模型决策树深度学习运动行为课堂行为策略研究关系研究 learning performance prediction model decision tree deep learning physical exercise behavior classroom behavior strategy research relationship research

分类号 G40-057 [文化科学—教育学原理]