Increasingly, publishers create and professors use online learning resources to help students master their coursework.  Cengage Publishing’s online learning system, MindTap, provides such resources and additionally collects data on student engagement with the online material for each course. This research applies data mining techniques, frequently dubbed ‘analytics’ in nowadays usage, to that data with the goal of predicting student achievement on exams given in the course. After contrasting traditional statistical methods with modern analytics practices, it specifically employs linear regression, neural networks, and regression trees to analyze the data with the intent of predicting student exam scores. Following the recommended method of an analytics approach, the paper employs a multi-model strategy, a so-called ensemble, for prediction purposes.