The Predict Model

Blackboard Predict uses established methodologies to identify historical patterns unique to your institution. Our algorithms identify which features best predict student risk at your institution. We use those features to create a model, or a set of assumptions from data, that displays in Predict. To ensure the highest quality, the Predict model is evaluated through statistical methods and a human review process.

What data is included in the model?

Our model includes the following information on your institution and each student:

  • Each student’s current activity and grades in the course, as well as performance relative to peers.
  • Course design factors such as instruction method, course level, or course size.
  • Each student’s past performance at your institution, both overall and in related courses.
  • The difficulty of a course compared to ones each student has already completed.
  • How others in each student’s major perform in the courses with the same subject.
  • Each student’s demographic and financial aid information.

How does the model work in a course?

Our model adapts throughout the length of your course. Initially, risk scores are factored are demographic information and past student performance. Each student’s grade and course activity will become a factor as the course progresses and more grades are finalized. By the end of the course, grade and course engagement are weighted most heavily.

The model recognizes when grade and activity become reliable indicators of the final course grade. This allows instructors to identify at-risk students early on, creating the opportunity to intervene before a student’s potential poor performance.