Modeling

Use ML when the logic is too complex to hand-code, patterns shift over time, or rules would be unmanageably numerous; use rules when the logic is simple, stable, and fully known.

Learning a mapping from inputs to outputs using a dataset of labeled examples (each input has a known target).

Finding structure or patterns in data that has no labels, e.g. clustering or dimensionality reduction.

An agent learns to take actions in an environment to maximize cumulative reward through trial-and-error feedback.

Supervised uses labeled data; unsupervised uses unlabeled data; reinforcement uses reward signals from an environment.