Predicting the learner's personality from educational data using supervised learning

Abstract

Differences in the learners' personality have an impact on their learning outcomes and achievements. Therefore, there is a need to automatically predict and identify their personalities in an unobtrusive way, and build the learner model accordingly. In this paper, we try to identify the learner's personality dimensions, according to the big five personality model, using educational data features in order to develop an automatic classifier that predicts the learner's personality discreetly based on his/her traces in an online learning system. We applied seven different supervised learning classification algorithms, using personality scores for each dimension (high or low) as target values, and analyzed the results. The findings were encouraging and revealed that most of Big Five personality dimensions can in fact be predicted using mainly educational data features, which could have an added value on unobtrusive dynamic learner modelling.