Learning analytics (LA) is a very hot topic that can be used to assess academic progress and predict future performance. The latest Horizon Report suggests that the time-to-adoption of LA by educational institutions will be around two to three years [1]. Thus, a lot of research and development initiatives are currently underway.

This interesting and well-written paper demonstrates a predictive analytic model for identifying students who are in danger of failing a course. It was created for the needs of the University of Phoenix.

Barber and Sharkey begin with a discussion of various known reasons students fail to graduate, as well as indicators that help identify a student that may have a tendency to fail. The authors show how they fine-tuned their model through three versions over several runs by adding variables. As a result, the Model 2 version accurately predicted 85 percent of outcomes for all students at week 0 (compared to 50 percent in Model 1).

The authors consider how LA can help teachers, educational managers, and students to predict course failure. LA can also be seen from other perspectives. For example, the outcome of this process can help instructional designers to better measure the quality of a course design and understand what works and what does not work [2]. In addition, LA can improve assessment of student performance by analyzing various indicators such as student postings and grades on assignments [3].

Adopting LA techniques is not easy. Stakeholders need usable analysis tools such as SNAPP [4] and NodeXL [5].

The development of new models like this one, as well as new techniques and tools, will undoubtedly help stakeholders analyze and interpret data about student progress and learning behavior.