Predict a Student's Major from Course Grades

# Predict a Student's Major from Course Grades You receive records of courses and grades taken by each student across academic years. Design a model that predicts the student's major. Explain how you would construct examples and features, choose baselines and models, evaluate generalization, and communicate limitations. ### Constraints & Assumptions - Students have variable-length histories, missing grades, and different course opportunities. - Features recorded after a major is declared may reveal the label rather than predict it. - The prediction must not be presented as certainty or used to constrain student choice. ### Clarifying Questions to Ask - At what academic point is the prediction made? - How are undeclared, changed, or multiple majors labeled? - Must the model work for future cohorts and new courses? ```hint Build examples at one consistent cutoff The same student history can be valid or leaked depending on when the prediction is supposed to occur. ``` ### What a Strong Answer Covers - Leakage-safe cohort and target construction. - Sparse transcript features, baselines, and justified model complexity. - Student-level and time-aware splits with imbalance-aware metrics. - Calibration, subgroup review, drift, interpretability, and responsible use. ### Follow-up Questions - How would you update the model when the course catalog changes? - What would you do if performance is high only because of required major courses?