The Summer & Winter Programs at UConn has been contributing to the academic success of students throug hundreds of courses offered in both intersession periods. Revenue from the program has been funding academic work and infrastructure from the University’s core to the many departments, programs, and campuses who participate in the program.
Scenario: The university is interested in knowing the probability a student is likely to enroll in the next Winter or Summer sessions, as well as factors that drive student retention.
Data: We use student performance and biographic data, as well as historical participation in the program. Data is kept on our staging Oracle Database, as well as in our analytic development environment. Data is versioned.
Modeling: We use a number of algorithms and pipelines, including XgBoost, Gradient Boosting, Voting Classifier, SGD, etc. and search for the best performing pipeline. Models are versioned.
Deployment: The model is exported to be used in WebFocus. Reports and Charts and Dashboards are created in WebFocus. The model can be used via Web Service (if requested).