• Worked in a team of 3 to design ML models such as kNN, IRT, matrix factorization, neural networks, to predict a student’s performance on a question given the students’ question-answering history.

• Achieved a maximum test accuracy of 70.6% (using IRT).

• Modified kNN by also weighing information such as age difference and similarity between a question and commonly answered questions between two students. Improved accuracy to 78.4%