Leveraging Predictive Analytics in the Cloud: A New Era for Student Success

Session Abstract: In this session, we will share our experience of transforming a student-retention predictive model from a vendor-supported model to an internally developed model using cutting edge cloud computing resources. We will focus on how newly available tools and technologies in cloud computing platforms can be successfully leveraged to develop and test student-success prediction modeling, and some of the challenges and pitfalls we encountered during our implementation. These include developing the necessary integrations and ingestions from source systems into our data lake and database development to preprocess and stage data for model training, feature engineering, model specification, data visualization, and dashboard development. We will also share input we have received from our student case-management team and associate deans who served as beta testers of the dashboards. Finally, we will discuss how we plan to leverage the same process for additional predictive models going forward.

Keywords: predictive analytics, predictive models, machine learning, student success, student retention