These notes represent my own study jottings and learning synthesis, not an authoritative or foolproof standard. If you find them helpful, feel free to use them. If you spot anything that could be clearer or have suggestions for additional concepts to include, issues and pull requests are warmly welcome.
Classification
Categorical decisions, probabilities, and decision boundaries
04. Classification
Discrete decisions, probabilistic outputs, and the geometry of decision boundaries.
Classification is the task of choosing among categories. That sounds simpler than regression, but most of the difficulty moves into two places: the probability model that turns a score into a belief, and the threshold that turns a belief into an action.
This chapter covers both. It starts with the shape of the problem, builds logistic regression from a linear score and a Bernoulli likelihood, examines the sigmoid that connects them, and then looks at what the resulting decision boundary actually looks like.