云计算下基于认知的学习质量评价优化算法的研究
发布时间：2018-10-13 19:41
【摘要】：本文的选题来自辽宁省教育厅人文社会科学“移动云计算环境下基于认知模型的高校教学评价与优化模型”的项目。研究的出发点是为了在云计算环境下建立行之有效的辅助教学效果的评价模型。本文主要对其中的学习质量评价方面的问题进行深入研究。评价系统平台的结构设计从评价系统的工作流程出发,将平台分为四个模块分别为:数据采集模块、数据存储模块、数据分析模块和数据查询模块。通过数据采集模块进行数据的收集,经过数据存储模块进行数据存储,为之后的数据分析模块做数据准备,数据分析之后,用户可以通过数据查询模块查询该用户的预测结果。其中最为重要的是数据分析模块,在该模块中通过改进算法的方式使预测效果达到更佳。本研究主要分为两方面:一方面是调查问卷题项的选取,另一方面是回归预测的算法优化。在进行《基于认知的学习质量评价调查问卷》的编制时,从认知学习理论出发,采用经过国际公认的《学习动机策略问卷》(MSLQ)为基础,并结合《学习质量评价:SOLO分类理论(可观察的学习成果结构)》一书修改得来。将首轮获得的调查数据使用SPSS进行效度和信度的检验,剔除掉不适用的题项,保留对学习质量影响较大的题项。经过内部一致性的检验由之前的0.926提升到0.930。说明问卷的可靠性较高,能够作为基于认知的学习质量评价的依据。用于回归预测的数据是通过调查问卷的再次发放回收得来。在云计算平台上使用Python语言编写相应的算法,对数值型数据进行回归预测。回归预测时将会面临两个问题,一是欠拟合、二是扩展瓶颈。本研究要通过多个自变量预测因变量,所以首先采用多元线性回归算法预测。进行回归预测的目的是想得到更佳的预测效果,由于多元线性回归算法确定的回归模型是要满足所有样本的规律,往往会把异常点也考虑到模型中,最终得到的模型将会出现欠拟合的现象。为了能够提高预测效果,改进基本算法给局部的点加权,采用局部加权线性回归算法提高预测的准确度。随着研究的深入,数据量将会不断的激增,在进行数据分析时将会出现扩展瓶颈的问题,为了能够让改进的算法在运算速率上有所提升,将局部加权线性回归算法进行并行化的处理,使其符合MapReduce编程模型的运算要求。最后通过数据分析得到的复相关系数R和残差的平方和的大小来确定模型的精度。用户通过本评价系统预测学习质量,便于用户及时对之后的学习做合理规划。
[Abstract]:The topic of this paper comes from the project of Humanities and Social Sciences of Liaoning Provincial Education Department, "Evaluation and Optimization Model of College Teaching based on Cognitive Model in Mobile Cloud Computing Environment". The starting point of the research is to establish an effective evaluation model of assistant teaching effect in cloud computing environment. This paper mainly studies the problems of learning quality evaluation. Based on the workflow of the evaluation system, the platform is divided into four modules: data acquisition module, data storage module, data analysis module and data query module. Data collection module through the data storage module for the subsequent data analysis module to do data preparation after the data analysis the user can query the user's prediction results through the data query module. The most important one is the data analysis module, in which the prediction effect is better by improving the algorithm. This research is mainly divided into two aspects: on the one hand, the selection of questionnaire items, on the other hand, the optimization of regression prediction algorithm. On the basis of cognitive learning theory and the internationally recognized Learning motivation Strategy questionnaire (MSLQ), a cognitive learning quality evaluation questionnaire was compiled. It is revised by the book Evaluation of Learning quality: SOLO Classification Theory (observable structure of Learning results). SPSS was used to test the validity and reliability of the data obtained in the first round of the survey. The inapplicable items were eliminated and the items that had a great impact on the learning quality were retained. After the internal consistency test from the previous 0.926 to 0. 930. It shows that the reliability of the questionnaire is high and can be used as the basis of cognitive learning quality evaluation. The data used for regression forecasting are recovered through the re-distribution of questionnaires. In the cloud computing platform, Python language is used to program the corresponding algorithm to predict the numerical data. Regression prediction will be faced with two problems, one is under-fitting, the other is expansion bottleneck. In this study, multiple independent variables are used to predict dependent variables, so multivariate linear regression algorithm is first used to predict dependent variables. The purpose of regression prediction is to get better prediction effect. Because the regression model determined by multivariate linear regression algorithm is to satisfy the laws of all samples, the outliers are often taken into account in the model. The resulting model will be underfitted. In order to improve the prediction effect, the local weighted linear regression algorithm is used to improve the accuracy of prediction. With the deepening of the research, the amount of data will continue to surge, and the problem of expansion bottleneck will occur in the analysis of the data, in order to improve the speed of the improved algorithm. The local weighted linear regression algorithm is parallelized to meet the operational requirements of the MapReduce programming model. Finally, the precision of the model is determined by analyzing the complex correlation coefficient R and the squared sum of the residuals. Users predict the learning quality through this evaluation system, which is convenient for users to make reasonable planning for the later learning in time.
【学位授予单位】：沈阳师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：G642.0;O213
本文编号：2269694
[Abstract]:The topic of this paper comes from the project of Humanities and Social Sciences of Liaoning Provincial Education Department, "Evaluation and Optimization Model of College Teaching based on Cognitive Model in Mobile Cloud Computing Environment". The starting point of the research is to establish an effective evaluation model of assistant teaching effect in cloud computing environment. This paper mainly studies the problems of learning quality evaluation. Based on the workflow of the evaluation system, the platform is divided into four modules: data acquisition module, data storage module, data analysis module and data query module. Data collection module through the data storage module for the subsequent data analysis module to do data preparation after the data analysis the user can query the user's prediction results through the data query module. The most important one is the data analysis module, in which the prediction effect is better by improving the algorithm. This research is mainly divided into two aspects: on the one hand, the selection of questionnaire items, on the other hand, the optimization of regression prediction algorithm. On the basis of cognitive learning theory and the internationally recognized Learning motivation Strategy questionnaire (MSLQ), a cognitive learning quality evaluation questionnaire was compiled. It is revised by the book Evaluation of Learning quality: SOLO Classification Theory (observable structure of Learning results). SPSS was used to test the validity and reliability of the data obtained in the first round of the survey. The inapplicable items were eliminated and the items that had a great impact on the learning quality were retained. After the internal consistency test from the previous 0.926 to 0. 930. It shows that the reliability of the questionnaire is high and can be used as the basis of cognitive learning quality evaluation. The data used for regression forecasting are recovered through the re-distribution of questionnaires. In the cloud computing platform, Python language is used to program the corresponding algorithm to predict the numerical data. Regression prediction will be faced with two problems, one is under-fitting, the other is expansion bottleneck. In this study, multiple independent variables are used to predict dependent variables, so multivariate linear regression algorithm is first used to predict dependent variables. The purpose of regression prediction is to get better prediction effect. Because the regression model determined by multivariate linear regression algorithm is to satisfy the laws of all samples, the outliers are often taken into account in the model. The resulting model will be underfitted. In order to improve the prediction effect, the local weighted linear regression algorithm is used to improve the accuracy of prediction. With the deepening of the research, the amount of data will continue to surge, and the problem of expansion bottleneck will occur in the analysis of the data, in order to improve the speed of the improved algorithm. The local weighted linear regression algorithm is parallelized to meet the operational requirements of the MapReduce programming model. Finally, the precision of the model is determined by analyzing the complex correlation coefficient R and the squared sum of the residuals. Users predict the learning quality through this evaluation system, which is convenient for users to make reasonable planning for the later learning in time.
【学位授予单位】：沈阳师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：G642.0;O213
【参考文献】
相关期刊论文 前10条
1 解峰;;运用元认知策略提高高职英语专业学生学业水平——基于语言学科平台的研究[J];山西青年职业学院学报;2016年02期
2 孙丽娜;杨殊;;学习质量评价中数据采集系统的设计与实现[J];软件导刊(教育技术);2016年02期
3 王新刚;孔云峰;;城市住房价格局部线性地理加权回归分析——以湖北省黄石市为例[J];中国土地科学;2015年03期
4 颜春宁;;局部优化加权回归算法在电力设备失效预测中的应用[J];计算机测量与控制;2014年01期
5 陈学军;黄利华;;基于云计算的义务教育学科课程资源共建共享模式[J];中国电化教育;2013年01期
6 张怀南;杨成;;我国云计算教育应用的研究综述[J];中国远程教育;2013年01期
7 赵子云;左明章;邓果;;基于云计算的教育信息公共服务平台的构建[J];现代教育技术;2012年12期
8 王丽娜;杨亭亭;刘仁坤;;国内外高等教育学习评价现状研究综述——兼论对国家开放大学学习评价体系建设的启示[J];现代远距离教育;2012年02期
9 包文婷;李三福;丁丽;;农村小学生学习动机、学习态度及其学习质量的实证研究[J];当代教育论坛(管理研究);2011年11期
10 张芝萍;曹燕华;郑栋;俞扬;;基于AHP应用的多元化、开放式高职学生学习质量评价体系研究[J];浙江师范大学学报(社会科学版);2011年06期
相关博士学位论文 前1条
1 王文婧;移动云计算的QoE评价与优化研究[D];北京邮电大学;2013年
相关硕士学位论文 前7条
1 万伟权;放疗中肿瘤运动基于实时跟踪呼吸预测的算法研究[D];南方医科大学;2015年
2 周章海;基于云计算理念的职业教育园区资源共享研究[D];大连海事大学;2014年
3 李志强;云计算教育服务平台模型分析和应用研究[D];广东技术师范学院;2013年
4 邓丽博;高中学生立体几何学习质量评价研究[D];湖南师范大学;2013年
5 杜玲玲;基于MapReduce的数据挖掘算法研究与应用[D];桂林电子科技大学;2012年
6 黄静;初中生数学学习兴趣、自我效能感、学业情绪与数学学业成绩的关系研究[D];四川师范大学;2012年
7 姜文;基于Hadoop平台的数据分析和应用[D];北京邮电大学;2011年
，本文编号：2269694
本文链接：https://www.wllwen.com/jiaoyulunwen/gaodengjiaoyulunwen/2269694.html
