Predictive Modeling & RioPACE

Too many online students were struggling, and our support systems were reactive rather than proactive. We often didn’t know a student was at risk until they had already disappeared. That had to change.

By 2008, Rio Salado College was encountering a persistent challenge: too many online students were struggling, and our support systems were reactive rather than proactive. We often didn’t know a student was at risk until they had already disappeared. That had to change.

What We Did

As Vice President of Academic Affairs, I led the effort to design a system that would provide early, actionable insight into student engagement and allow faculty to intervene before it was too late. The goal was ambitious—create a predictive model integrated into our LMS that recalibrated weekly and could scale across hundreds of courses.

We began with a pilot using naïve Bayes modeling and over 30 variables—logins, click patterns, pacing, prior academic behavior—and tested it in fifteen high-volume courses. Encouraged by the results, I spearheaded the creation of RioPACE: a Progress and Course Engagement engine that assigned weekly risk levels to each student. The system was visual, intuitive, and embedded directly into the course environment.

Result

More than 90% of online undergraduate courses adopted RioPACE, and faculty were empowered with real-time insights to reach struggling students early. The result was improved course completion, stronger student engagement, and national recognition for innovation in predictive analytics.