Predictive Models for Students: Ethical and Practical Considerations

Machine learning algorithms are capable of accurately predicting student outcomes. However, ethical and practical decisions must be made about what variables should be included in these models. In this session, we will compare the accuracy and impact of predictive models using different feature sets and explore how institutions might best leverage these predictions.

Outcomes: Understand the differences in accuracy across the semester of predictive models using different feature sets and the different ways models can be "wrong" * Identify the potential benefits and consequences of including certain features * Explore when and how predictions might best be shared with students