Predicting Student Persistence Using Machine Learning: An End-to-End Overview of Local Development to Production Deployment

As part of the Office of University Provost, Actionable Analytics team developed machine-learning models to predict next-term persistence. Towards this effort, we iteratively constructed a list of features for each student and built a set of predictive models to map these features to a predicted probability. This presentation will cover the end-to-end development to deployment of these models to inform decision making and student-level actionability aimed at increasing student success.