2026 ASEE Annual Conference & Exposition

Mastery learning is based on the theory that nearly all students can succeed in a course given sufficient time and appropriate feedback and practice opportunities. It has been used in many disciplines to allow students to engage in the learning cycle until they demonstrate mastery. We use data from a custom mastery learning platform to predict which students will struggle in a graduate introductory programming course. Data from 100 autograded formative programming exercises, including grade, date graded, whether passed, timeliness, and number of submissions, are analyzed together with summative student performance on three tests, a final exam, and an overall course grade. Results include a rule-based model and machine learning model for predicting student performance, as well as practical recommendations for research-based mastery learning policies. A database tool is proposed to monitor student performance on formative tasks and alert them when a change of approach is advised.