Capture Enrollment Predictions 98% Accurate Across Partners

Once again, the enrollment predictions generated by Envision, Capture Higher Ed’s next-generation predictive engine for higher education, have proven to be highly accurate. For the second straight year, our final predictions were 98% accurate across all partners. For 47% of Capture clients that use Envision, our predictions were within 10 students.

Envision is the only comprehensive predictive engine in higher education, utilizing machine learning and a non-linear analytics tool set to predict with unprecedented accuracy who will apply and who will enroll.

The 98% represents the last prediction schools asked us to make for them, usually after their deposit deadline. However, we were still 96% accurate in April — before many students had deposited. Overall, we were within 2.5 percentage points in our predictions from November through March. From April through July, our predictions are mostly within 1.5 percentage points.

We start with good predictions that only get better as we go through the enrollment cycle.

Getting Better

Also, the way we make predictions keeps getting better. It’s hard to improve upon 98% accuracy, but we can get there sooner. Capture data scientists have worked hard for months to improve Envision’s accuracy even further and we’ve made some improvements to the way we make predictions for fall 2019. This includes taking into account institutional policy decisions and dialing in the way we weight deposits — that should enable us to zero in on the correct prediction earlier in the cycle.

By John Foster, Senior Data Analyst, Capture Higher Ed