We use the results from tests, quizzes, in-class questionnaires, games and user interactions to create a model for how a user might best learn. This model is then used to predict where a user will next have trouble, which we then remedy with extra tutoring.