Using predictive analytics for at-risk students at University of North Carolina
How might we invest in a robust predictive analytics system that maximizes support for our highest need students?
- Developed a predictive analytics model to determine likelihood of student attrition and success based on academic, economic, and social factors
- Developed a tool to first identify students likely to struggle academically; then deployed relevant data in decision-making (e.g., advisors could auto-enroll incoming students in a student success course)
- Offered at-risk/high-need students support (e.g., TRIO Student Support Services and Frontier Set Summer Bridge program)
The Vice Chancellor of Enrollment Management developed UNC-Greensboro’s predictive analytics model using Rapid Insight. Key success indicators for a new first-time, first-degree cohort include high school performance, hours and types of credits enrolled, declared major, times between key dates, family involvement, as well as parent/guardian academic experience, geographic location, and affordability. The model places students into deciles 1-10, and staff uses this information to reach out to at-risk students (deciles 1-4) with specialized support services.