Classification Methods

This lecture introduces three baseline classifiers: logistic regression (linear score mapped to probability), k-nearest neighbors (let nearby examples vote), and naive Bayes (Bayes’ rule with independence assumption). We discuss how to evaluate classifiers beyond accuracy using confusion matrices, precision, recall, and F1 score, and how to compare classifiers fairly under the same resampling rule.

Interactive: Decision boundaries

Different classifiers produce different decision boundaries on the same data. Logistic regression draws a straight line; k-NN produces a flexible boundary; decision trees create axis-aligned splits.