ABSTRACT
This chapter shows how key metrics are being derived from the data by researchers and outlines three statistical methods commonly used in learning analytics: linear regression, logistic regression and naive Bayes. Simple metrics might include the number of times students have logged into the learning management system (LMS) or the number of forum postings they have made. A composite metric that combines several metrics is likely to be created in order to decide when to intervene in some way with students. Tabaa & Medouri developed a system for analysing the big data accumulated from massive open online courses (MOOCs) to identify learners who are likely to drop out. Meanwhile, in a course where use of the forum is regarded as essential, a metric that measures forum usage is likely to have more relevance for predicting student attainment than in one where online discussion is optional. One of the most commonly used methods for predictive analytics is linear regression.
