Mining Online Interaction Traces for Student Success Prediction in Flipped Classrooms
Increasingly adopted over the years, flipped classrooms represent a learning design that requires students to complete pre-class learning activities before participating in face-to-face sessions. For pre-class activities, this design often makes use of videos and digital content published in online platforms. The students’ engagement in pre-class activities is essential for the success of flipped classrooms, as these activities prepare students for effective participation in face-to-face sessions.
The widespread adoption of a flipped classroom design has spurred investigations into the issues of how to anticipate academic performance by analyzing digital traces, gathered from students’ interactions along pre-class activities. Student success prediction, where a model forecasts future performance of students as they interact online, is a primary while challenging goal. Trustworthy early predictions enable effective content personalization and adaptive teaching interventions.
This project aims at exploring behavioral patterns and predicting students’ future performance in flipped-classroom settings. As a case study, we consider a Linear Algebra course part of the EPFL Bachelor’s Programs, delivered by Prof. Simone Deparis under a flipped-classroom approach in the EPFL Courseware Platform, since 2017. The extent to which success at the end of the course can be anticipated is analyzed by examining interaction traces left by students while interacting with content (e.g., videos), peers (e.g., forums), and assessments (e.g., quizzes). With the support of the Center for Digital Education (CEDE), the Center for Learning Sciences (LEARN), and the Teaching Support Center (CAPE), several indicators of engagement in online pre-class activities are being analyzed as predictors of students’ performance, and the resulting machine-learning models with these indicators as features are being developed.
The contributions coming from this project can shape intelligent learning platforms able to provide a data-driven formative feedback. These insights are essential to assist students in regulating their use of online learning resources and inform teachers on when, where and why to intervene.