Characterizing Student Engagement Moods for Dropout Prediction in Question Pool Websites

Problem-Based Learning (PBL) is a popular approach to instruction that supports students get hands-on training by solving problems. Question Pool websites (QPs) such as LeetCode, Code Chef, and Math Playground help PBL supplying authentic, diverse, contextualized questions students. Nonetheless, empirical findings suggest 40% 80% of registered in QPs drop out less than two months. This research the first attempt understand predict student dropouts from via exploiting students' engagement moods. Adopting data-driven approach, we identify five different moods for QP students, which are namely challenge-seeker, subject-seeker, interest-seeker, joy-seeker, non-seeker. We find have collective preferences answering each mood, deviation those increases their probability dropping significantly. Last but not least, this paper contributes introducing new hybrid machine learning model (we call Dropout-Plus) predicting QPs. The test results on China, with nearly 10K show Dropout-Plus can exceed rival algorithms' dropout prediction performance terms accuracy, F1-measure, AUC. wrap up our work giving some design suggestions managers online professionals reduce dropouts.