Student Performance Prediction by Discovering Inter-Activity Relations
Student Performance Prediction by Discovering Inter-Activity Relations
复制标题
通过发现活动间关系来预测学生表现
DOI：
--
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
Peter Brusilovsky
中科院分区：
文献类型：
--
作者：
Shaghayegh Sherry Sahebi;Peter Brusilovsky
© 2018国际教育数据挖掘协会。All rights reserved.性能预测已经成为利用大量在线学习数据的最流行的方法之一。在现有的研究中，成绩预测主要基于学生过去在分级学习资源（如问题和测验）中的活动，而忽略了学生在非分级学习资源（如阅读材料）中的活动。本文介绍了一种利用学生在非分级学习资源中的作业作为辅助数据，预测学生在分级学习资源中的表现的方法。该方法仅利用学生活动数据就能发现不同类型学习资源之间隐藏的相互关系。基于我们的实验，所提出的方法可以显着减少学生成绩预测的错误，相比基线算法，同时发现有意义的和令人惊讶的学习资源之间的关系。
© 2018 International Educational Data Mining Society. All rights reserved. Performance prediction has emerged as one of the most popular approaches to leverage large volume of online learning data. In the majority of current works, performance prediction is based on students’ past activities in graded learning resources (such as problems and quizzes), while their activities in non-graded resources (such as reading material) are ignored. In this paper, we introduce an approach that can take advantage of students’ work with non-graded learning resources, as auxiliary data, in order to predict students’ performance in graded resources. This approach can discover the hidden inter-relationships between learning resources of different types, only using student activity data. Based on our experiments, the proposed approach can significantly reduce the error of student performance prediction, compared to baseline algorithms, while discovering meaningful and surprising relationships among learning resources.