PROGNOSING THE RESULTS OF STUDENTS' EDUCATIONAL ACTIVITIES

Efficiency prediction has emerged as one of the most popular ways to use large amounts of online educational data. In the majority of current work, efficiency prediction is based on students ' past activities in step learning resources (e.g. problems and quizzes), while their activities on unvalued resources (e.g. reading materials) are ignored. In this article, we provide an approach that can take advantage of students ' work with unrated learning resources as auxiliary data to predict the effectiveness of students in evaluated resources. This approach can only identify hidden correlations between different types of learning sources using student activity data. Based on our experiments, the recommended approach, while identifying meaningful and surprising relationships between learning resources, can significantly reduce the error in predicting student activity compared to elementary algorithms.