The analysis method consisted of two steps: run Expectation Maximization to fit the parameter on the training set for each model, and apply the trained parameters to the test sets to predict the student performance of each question. The motivation behind this method is to compare the overall performance of each model, including the original KT model, the KT-Forget model, and the KT-Slip model. They trained the proposed model on datasets that are collected from the real-world Intelligent Tutoring System – the Cognitive Tutor, and for further reference of the models’ results they also apply their models to the datasets that are collected from the ASSISTments Platform. They evaluated and compared the accuracy of the KT-slip, KT-forget and regular KT model by calculating Residual and Area Under Curve (AUC). Residual is the mean of the actual performance subtracted by the predicted performance. AUC is a robust accuracy measure where a score of 0.50 represents a model that is only as good as chance and 1.0 represents a perfectly predicting model.