基于Blackboard平台的在线学习行为分析与预测

发布时间：2018-06-21 04:29

本文选题：在线学习 + 学习行为分析　；参考：《内蒙古师范大学》2017年硕士论文

【摘要】：近年来,随着互联网与通信技术的发展,基于网络的在线学习以其灵活性、开放性的特点吸引着众多教育研究者和学习者。尽管网络学习以学习者为中心,打破时空限制,实现了学习者的自主学习和个性化学习,但是在线学习的时空分离限制了师生信息交流的直接性和及时性,教师无法直接对学习者的学习行为特征进行观察,也不利于教师及时给学习者提供有针对性的指导和帮助。学习分析技术能够提高网络教学质量和在线学习效果,为预测学生的在线学习效果、判定影响学习者成绩的关键因素提供了可行的方法。随着在线学习的发展,学习分析技术将成为未来教育技术研究的主要内容。本文在行为科学理论的基础上,结合Blackboard网络课程,以《心理学概论》网络课程的在线学习行为作为研究对象,根据相关网络学习行为模型,把在线学习行为分为五类,并将可作为学习者行为特征的数据进行归类。对五类网络学习行为分析,通过统计学的方法,确定影响成绩的数据变量,构建学习成绩预测模型并验证模型的可靠性。本论文共分为六个部分:第一部分是绪论部分,阐述了论文研究的背景、意义、国内外研究现状等内容。第二部分是理论基础及相关概念,介绍了行为科学理论,并对在线学习行为和学习分析技术进行了概念界定。第三部为网络课程简介与数据的获取,根据网络学习行为模型,针对Blackboard网络课程数据,参考学习者规范和对象元数据规范选取行为特征的量化参数。第四部分是在线学习行为分析,根据选取的网络行为特征的数据变量,对在线学习行为进行分析,在分析学习者活动规律的同时初步判定了与学习效果相关的学习行为。第五部分是成绩预测模型的建立与验证,首先从统计学的角度探究与学习效果有关的学习行为,然后对学习者的期末成绩建立回归模型,最后选取60分为临界值,使用二项逻辑回归分析验证成绩预测模型的准确率。第六部分为研究的结论与建议,研究发现与成绩呈显著相关的数据变量有13个,最终建立的模型解释度为29%,自变量在线天数、测试总分、题库点击量、发布博客的数量显著,二项逻辑回归方程预测准确率为78.1%,最后根据结论提出建议。本研究帮助教师及时了解学习者的在线学习情况提供了参考,对存在学习风险的学习者及时进行干预和帮助。帮助教师进行课程和资源的设计、开发,合理安排和组织网络课堂教学活动,并且制定网络课程考核标准。
[Abstract]:In recent years, with the development of Internet and communication technology, web-based online learning has attracted many educational researchers and learners for its flexibility and openness. Although online learning is learner-centered, breaks the limitation of time and space, and realizes learner's autonomous learning and individualized learning, the separation of online learning time and space limits the directness and timeliness of information exchange between teachers and students. Teachers can not directly observe the characteristics of learners' learning behavior, nor is it conducive for teachers to provide timely guidance and help to learners. Learning analysis technology can improve the quality of online teaching and the effect of online learning. It provides a feasible method for predicting students' online learning effect and judging the key factors that affect learners' achievement. With the development of online learning, learning analysis technology will become the main content of educational technology research in the future. On the basis of behavioral science theory and Blackboard network course, this paper takes the online learning behavior of the network course as the research object, according to the related network learning behavior model, divides the online learning behavior into five categories. The data can be classified as learner behavior characteristics. In the analysis of five kinds of online learning behaviors, the data variables that affect the achievement are determined by statistical method, and the prediction model of learning achievement is constructed and the reliability of the model is verified. This paper is divided into six parts: the first part is the introduction part, which describes the background, significance, research status at home and abroad. The second part is the theoretical basis and related concepts, introduces the behavioral science theory, and defines the online learning behavior and learning analysis technology. The third part is the introduction of online courses and the acquisition of data. According to the online learning behavior model, the quantitative parameters of behavior characteristics are selected according to the Blackboard online course data, referring to the learner specification and object metadata specification. The fourth part is the analysis of online learning behavior. According to the selected data variables of network behavior characteristics, the online learning behavior is analyzed, and the learning behaviors related to learning effect are preliminarily determined while analyzing the rules of learners' activities. The fifth part is the establishment and verification of the achievement prediction model. Firstly, it explores the learning behaviors related to the learning effect from the perspective of statistics, then establishes a regression model for the final grades of the learners, and finally selects 60 as the critical value. The accuracy of the performance prediction model was verified by binary logistic regression analysis. The sixth part is the conclusions and recommendations of the study. The study found that there are 13 data variables significantly related to the results, the final interpretation of the model is 29, independent variables online days, total test scores, number of hits to the question bank, the number of blog posts is significant. The prediction accuracy of binomial logistic regression equation is 78.1. Finally, some suggestions are put forward according to the conclusion. This study provides a reference for teachers to understand learners' online learning in time, and provides timely intervention and help for learners with learning risks. To help teachers design, develop, arrange and organize the online classroom teaching activities, and establish the assessment standard of online courses.
【学位授予单位】：内蒙古师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：G434;B84-4

【参考文献】