Personalized Multi-Regression Models for Predicting Students' Performance in Course Activities

Asmaa Elbadrawy, R. Scott Studham, and George Karypis
UMN CS 14-011, 2014
Download Paper
Methods that accurately predict the grade of a student at a given activity and/or course can identify students that are at risk in failing a course and allow their educational institution to take corrective actions. Though a number of approaches have been developed for building such performance prediction models, they either estimate a single model for all students based on their past course performance and interactions with learning management systems (LMS), or estimate student-specific models that do not take into account LMS interactions; thus, failing to exploit fine-grain information related to a student’s engagement and effort in a course. In this work we present a class of linear multi-regression models that are designed to produce models that are personalized to each student and also take into account a large number of features that relate to a student’s past performance, course characteristics, and student’s engagement and effort. These models estimate a small number of regression models that are shared across the different students along with student- specific linear combination functions to facilitate personalization. Our experimental evaluation on a large set of students, courses, and activities shows that these models are capable of improving the performance prediction accuracy by over 20%. In addition, we show that by analyzing the estimated models along with the student-specific combination functions we can gain insights on the effectiveness of the educational material that is made available at the courses of different departments.
Also appears in the 5th International Learning Analytics & Knowledge conference (LAK15), 2015.
Research topics: Data mining | Learning analytics