A course agnostic approach to predicting student success from VLE log data using recurrent neural networks

We describe a method of improving the accuracy of a learning analytics system through the application of a Recurrent Neural Network over all students in a University, regardless of course. Our target is to discover how well a student will do in a class given their interaction with a virtual learning environment. We show how this method performs well when we want to predict how well students will do, even if we do not have a model trained based on their specific course.