University of Michigan Furthers On-Campus Analytics with Help from Unizin Partnership

people in meeting, silhouetted against the sun

When Chris Brooks attended the Unizin Innovation Summit last October, he was hoping to walk away with more knowledge about Unizin. He found more than he expected.

Brooks, a Research Assistant Professor in the School of Information and Director of Learning Analytics and Research at the Office of Academic Innovation, was in the midst of adapting an early warning system for faculty at the University of Michigan to leverage Canvas data. The goal was to build predictive at-risk models for students that would notify faculty early in the term of student progress compared to expected outcomes. This would trigger prompts and help faculty decide whether to intervene.

At the time of the Summit, Brooks was trying to figure out how to best create models across the Unizin Canvas-based Data Warehouse and the on premise local data warehouse used at the University of Michigan in order to form a more complete picture of the data.

“We needed a tool that would allow to quickly join educational datasets across warehouses for research,” said Brooks. “While we were at the Summit we talked with Matthew Gunkel from the Learning Technologies Division of the University Information Technology Services at Indiana University, who had just gone through a similar process. They suggested a tool for data virtualization that dealt with our problem, allowing us to more rapidly integrate data from vendors with research data. At Unizin, we’re learning faster together about these new technologies.”

Indiana University provided the University of Michigan with examples of how data virtualization tool worked, advised them on the size and scope of the tool, and connected them with the appropriate contacts for the vendor.

As the University of Michigan builds out their capacity to map across data sources, Brooks and his team are interested in seeing which other institutions might be interested in a shared architecture that can be tested across institutions.

“There is an interesting opportunity to work across Unizin institutions,” said Brooks. “By building some predictive models we can look at [for example] an introductory Psychology class at multiple institutions and compare how predictive models might differ.”

Brooks is also working with Unizin’s Principal Data Scientist, J D Freeman, to explore comprehensive analytics solutions for use on the University of Michigan campus.