The Sorting Hat: A Predictive Modeling-powered Admissions Yield Dashboard in Power BI

Undergraduate admissions is a highly uncertain process. UGA Institutional Research has developed a Power BI dashboard that gives users have the ability to admit applicants in “buckets” with different admissions criteria before seeing the predicted value of key metrics (e.g. yield rate, residency split, average test scores) for each bucket and the entire admitted class. The predictions are powered by a random forest machine learning model that outputs a chance to yield for each applicant, which is then fed into Power BI measures to extrapolate the metrics.

Presenter(s):

Zhen Zhang University of Georgia