Foundations
Classical models
Perceptrons, support vector machines, decision trees, random forests, gradient-boosted trees and Gaussian classifiers. Gradient-boosted trees are still the usual first choice for tabular data, and margins, impurity and bagging come back in deep learning.