Early Prediction of Students at Risk of Failing a Face-to-Face Course in Power Electronic Systems

Early warning systems (EWSs) have proven to be useful in identifying students at risk of failing both online and conventional courses. Although some general reported acceptable ability work modules with different characteristics, those designed from a course-specific perspective recently provided better outcomes. Hence, the main goal this article is design tailored EWS for course power electronic circuits. For that purpose, effectiveness common classifiers predicting at-risk has been analyzed. slight differences their performance only noticed, an ensemble classifier combining outputs several them best performer. As major contribution, novel weighted voting combination strategy proposed exploit global information about how basic prediction algorithms perform on time points during semester diverse subsets student-related features. Predictions five critical analyzed, revealing end fourth week optimal identify course. At moment, accuracies 85%–90% reached. Moreover, scenarios attributes considered every point. Besides parameters student's background continuous assessment, features estimating progression weekly assignments introduced. The proposal set new input variables another key because they allowed improve more than 5% predictions