Video Discussion Week 3 Handout
EDUC 5183: Big Data, Education, and Society
Summer 2024
Professor Ryan Baker

• Which model is most accurate at predicting future performance in the learning system?
• Which model is most accurate at predicting future performance on external tests?
• Does a model say a student has mastered too quickly? (Leading to a student moving on who is not ready)
• Does a model say a student has mastered too slowly? (Leading to a student wasting time)
• Does a model display other weird behavior (such as lowering a knowledge estimate after a correct answer)?
• How much data does a model need in order to function correctly?
• Can a model take into account situations where an item involves multiple skills?
• How much should a model adjust itself to an individual student? (rate of learning, amount of guessing, etc.)