Clustering Students in ASSISTments: Exploring System- and School-Level Traits to Advance Personalization
Clustering Students in ASSISTments: Exploring System- and School-Level Traits to Advance Personalization
复制标题
将学生聚集在 ASSISTments 中：探索系统和学校层面的特征以促进个性化
DOI：
--
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
N. Heffernan
中科院分区：
文献类型：
--
作者：
Seth A. Adjei;Korinn S. Ostrow;E. Erickson;N. Heffernan
很少有人尝试创建学生模型，将学生和学校水平的特征聚集在一起，作为设计个性化学习干预措施的手段。在目前的工作中，ASSISTments的数据与公开的学校水平数据进行了丰富，并采用了K-Means聚类。结果揭示了学校地点、地区财富衡量和系统交互模式作为个性化潜在焦点的重要性。然后将聚类应用于一组保留数据的测试集，并使用聚类分配来帮助预测年终标准化数学考试成绩。研究结果表明，虽然聚类解释不能推广到保留数据，但聚类通常有助于预测标准化考试成绩
Few attempts have been made to create student models that cluster student and school level traits as a means to design personalized learning interventions. In the present work, data from ASSISTments was enriched with publicly available school level data and K-Means clustering was employed. Results revealed the importance of school locale, measures of district wealth, and system interaction patterns as potential foci for personalization. Clusters were then applied to a test set of held out data and cluster assignments were used to help predict end-of-year standardized mathematics test scores. Findings suggest that while cluster interpretations were not generalizable to held out data, clustering was generally helpful in predicting standardized test scores