学习测评大数据支撑下面向知识点的学习预警建模与仿真
|
摘要
学习测评是教学环节中的重要活动模块,对基于学生测评所产生的数据进行收集分析有助于挖掘学生真实的学习状态,明晰学生对知识点的掌握情况。以测评大数据为基础,以学生知识点综合掌握情况为研究内容,以贝叶斯网络为分析方法,通过节点权重分析及因果关系分析形成预警模型网络结构,在参数学习和条件概率分析的基础上形成面向知识点的学习预警模型。以某学校高一数学三次考试中的500条测评数据作为样本数据,采用5折交叉验证法对模型的有效性进行验证,采用因果推理和诊断推理对知识点预测结果进行仿真验证分析。研究结果表明,所构建的预警模型涵盖1个总目标层节点、5个指标层节点和20个数据层节点,通过调整模型中的节点能够有效观测不同条件下指标层和目标层知识点的掌握情况,并进行准确预警;通过灵敏度分析得出学习预警需要关注的核心节点,其影响程度从大到小依次为:核心知识点掌握情况>相关知识点掌握情况>各难度题目掌握情况;交叉使用主次指标排队分类法和专家打分法确定节点不同预警状态概率间的权重并以此计算判别值,按需调整其阈值以对学习者进行个性化类别划分与状态识别。
Learning assessment is an important activity module in the teaching process. The collection and analysis based on the data generated by the students' assessment helps to explore the real learning status of the students and clarify the students' mastery of the knowledge points. Based on the evaluation of big data,the study takes the comprehensive knowledge of students' knowledge points as the research content,and uses Bayesian network as the analysis method to form the network topology of early warning model through node weight analysis,which is formed on the basis of parameter learning and conditional probability analysis. Finally,500 data from three high school mathematics tests were used as sample data,and the validity of the model was verified by 5-fold cross-validation method. The results of knowledge point prediction were simulated and analyzed by causal reasoning and diagnostic reasoning. The research results show that the constructed early warning model covers one total target layer node,five index layer nodes and twenty data layer nodes. By adjusting the nodes in the model,it can effectively observe the knowledge of the index layer and the target layer under different conditions,and make an accurate warning; Sensitivity analysis leads to the core nodes that need to pay attention to learning and early warning. The degree of influence from large to small is as follows: mastery of core knowledge points > mastery of relevant knowledge points > mastery of difficult topics; Cross-use of primary and secondary indicators queuing classification method and expert scoring method to determine the weight between different early warning state probabilities of nodes and calculate the discriminant value,adjust its threshold according to the needs of learners to personalized requirements for classification and state recognition.
Learning assessment is an important activity module in the teaching process. The collection and analysis based on the data generated by the students' assessment helps to explore the real learning status of the students and clarify the students' mastery of the knowledge points. Based on the evaluation of big data,the study takes the comprehensive knowledge of students' knowledge points as the research content,and uses Bayesian network as the analysis method to form the network topology of early warning model through node weight analysis,which is formed on the basis of parameter learning and conditional probability analysis. Finally,500 data from three high school mathematics tests were used as sample data,and the validity of the model was verified by 5-fold cross-validation method. The results of knowledge point prediction were simulated and analyzed by causal reasoning and diagnostic reasoning. The research results show that the constructed early warning model covers one total target layer node,five index layer nodes and twenty data layer nodes. By adjusting the nodes in the model,it can effectively observe the knowledge of the index layer and the target layer under different conditions,and make an accurate warning; Sensitivity analysis leads to the core nodes that need to pay attention to learning and early warning. The degree of influence from large to small is as follows: mastery of core knowledge points > mastery of relevant knowledge points > mastery of difficult topics; Cross-use of primary and secondary indicators queuing classification method and expert scoring method to determine the weight between different early warning state probabilities of nodes and calculate the discriminant value,adjust its threshold according to the needs of learners to personalized requirements for classification and state recognition.
引文
[1]Pedro M S,Baker R S,Heffernan N.An Integrated Look at Middle School Engagement and Learning in a Digital Environment as Precursors to College Attendance[J].Technology Knowledge&Learning,2017,22(2):1-28.
[2]Vieira C,Parsons P,Byrd V.Visual learning analytics of educational data:A systematic literature review and research agenda[J].Computers&Education,2018,122(7):119-135.
[3]杨现民,唐斯斯,李冀红.发展教育大数据:内涵、价值和挑战[J].现代远程教育研究,2016(1):50-61.
[4]陈明选,许晓群,王玉家.基于教育测评数据分析的教学优化研究[J].中国电化教育,2018(5):80-89.
[5]Hu Y H,Lo C L,Shih S P.Developing early warning systems to predict students’online learning performance[J].Computers in Human Behavior,2014,369(7):469-478.
[6]Tempelaar D,Rienties B,Mittelmeier J,et al.Student profiling in a dispositional learning analytics application using formative assessment[J].Computers in Human Behavior,2017,78(8):408-420.
[7]Aguilar,Stephen J.Learning Analytics:at the Nexus of Big Data,Digital Innovation,and Social Justice in Education[J].Tech Trends,2017,62(1):37-45.
[8]Lee U J,Sbeglia G C,Ha M,et al.Clicker Score Trajectories and Concept Inventory Scores as Predictors for Early Warning Systems for Large STEM Classes[J].Journal of Science Education&Technology,2015,24(6):848-860.
[9]Pena-Ayala,Alejandro.Learning analytics:A glance of evolution,status,and trends according to a proposed taxonomy[J].Wiley Interdisciplinary Reviews:Data Mining and Knowledge Discovery,2018,8(3):98-110.
[10]肖巍,倪传斌,李锐.国外基于数据挖掘的学习预警研究:回顾与展望[J].中国远程教育,2018(2):70-78.
[11]Howard E,Meehan M,Parnell A.Contrasting Prediction Methods for Early Warning Systems at Undergraduate Level[J].The Internet and Higher Education,2017,(8):1-20
[12]Dabrowski J J,Beyers C,Villiers J P D.Systemic banking crisis early warning systems using dynamic Bayesian networks[M].Pergamon Press,Inc.2016.
[13]Xu J G,Zhao Y,Chen J,et al.A Structure Learning Algorithm for Bayesian Network Using Prior Knowledge[J].Journal of Computer Science and Technology,2015,30(4):713-724.
[14]邓雪,李家铭,曾浩健,等.层次分析法权重计算方法分析及其应用研究[J].数学的实践与认识,2012(7):93-100.
[15]陆静,王捷.基于贝叶斯网络的商业银行全面风险预警系统[J].系统工程理论与实践,2012(2):225-235.
[2]Vieira C,Parsons P,Byrd V.Visual learning analytics of educational data:A systematic literature review and research agenda[J].Computers&Education,2018,122(7):119-135.
[3]杨现民,唐斯斯,李冀红.发展教育大数据:内涵、价值和挑战[J].现代远程教育研究,2016(1):50-61.
[4]陈明选,许晓群,王玉家.基于教育测评数据分析的教学优化研究[J].中国电化教育,2018(5):80-89.
[5]Hu Y H,Lo C L,Shih S P.Developing early warning systems to predict students’online learning performance[J].Computers in Human Behavior,2014,369(7):469-478.
[6]Tempelaar D,Rienties B,Mittelmeier J,et al.Student profiling in a dispositional learning analytics application using formative assessment[J].Computers in Human Behavior,2017,78(8):408-420.
[7]Aguilar,Stephen J.Learning Analytics:at the Nexus of Big Data,Digital Innovation,and Social Justice in Education[J].Tech Trends,2017,62(1):37-45.
[8]Lee U J,Sbeglia G C,Ha M,et al.Clicker Score Trajectories and Concept Inventory Scores as Predictors for Early Warning Systems for Large STEM Classes[J].Journal of Science Education&Technology,2015,24(6):848-860.
[9]Pena-Ayala,Alejandro.Learning analytics:A glance of evolution,status,and trends according to a proposed taxonomy[J].Wiley Interdisciplinary Reviews:Data Mining and Knowledge Discovery,2018,8(3):98-110.
[10]肖巍,倪传斌,李锐.国外基于数据挖掘的学习预警研究:回顾与展望[J].中国远程教育,2018(2):70-78.
[11]Howard E,Meehan M,Parnell A.Contrasting Prediction Methods for Early Warning Systems at Undergraduate Level[J].The Internet and Higher Education,2017,(8):1-20
[12]Dabrowski J J,Beyers C,Villiers J P D.Systemic banking crisis early warning systems using dynamic Bayesian networks[M].Pergamon Press,Inc.2016.
[13]Xu J G,Zhao Y,Chen J,et al.A Structure Learning Algorithm for Bayesian Network Using Prior Knowledge[J].Journal of Computer Science and Technology,2015,30(4):713-724.
[14]邓雪,李家铭,曾浩健,等.层次分析法权重计算方法分析及其应用研究[J].数学的实践与认识,2012(7):93-100.
[15]陆静,王捷.基于贝叶斯网络的商业银行全面风险预警系统[J].系统工程理论与实践,2012(2):225-235.