Structure Discovery: Machine Learning for Behavioral Data

This lecture covers the concepts of Bayesian Knowledge Tracing (BKT), Generalized Linear Models, Item Response Theory, and Performance Factors Analysis. It delves into the Additive Factors Model (AFM) and Generalized Linear Mixed Effects Model, explaining how they are used in educational data mining. The instructor discusses the importance of tracing student knowledge and evaluating student models using metrics like RMSE and AUC. The lecture also explores clustering algorithms such as K-Means Clustering and Spectral Clustering, emphasizing the process of selecting the optimal number of clusters. Practical examples and methods for structure discovery in behavioral data are presented, including the initialization of cluster centers and the spectral clustering algorithm.