Predicting Student Success according to Online Activities in a Blended Course using Artificial Neural Networks

In a blended course, some portion of the classes is held as the traditional face-to-face approach whereas the rest is conducted as a web-based online learning approach. In this paper, we focus on a selected blended course in order to observe the effects of online activities on the final success of students in that course. We opt to calculate the success of a student in a course with two metrics. The first success metric is the letter grade and the second is if the student passes or fails. We would like to predict the student success by applying an artificial neural network (ANN) model. In the model, we provided different set of features from the collected features The experiment results indicate that predicting grade letters is much more prone to errors than predicting the pass/fall result. The tests show that we can predict if a student pass or fail with about 81% accuracy considering only the number of online activities. These results indicate that, in a blended course, student success is not only determined by the quantity of online activities but also it might be related with the quality of the face-to-face interaction with the instructor.