Learning Analytics (LA) consists of statistical models and algorithms for understanding the relationship between a set of learners and a set of test items. LA methods automate crucial educational tasks such as inferring which concepts a learner understands well, which ones they do not, and how these concepts relate to the individual test items. This information can be used to predict future learning outcomes as well as adapt instruction to achieve specific educational goals.
We adopt a fully Bayesian approach to LA and develop several novel LA methods which have superior flexibility, interpretability, and performance over state-of-the-art methods. We additionally develop methods that use LA to perform collaboration-type identification between learners, where we not only identify collaboration in a statistically principled way, but also classify the type of collaborative behavior employed.