Supervised, unsupervised, reinforcement

The three paradigms

• Supervised learning — you have labelled examples (input → known answer). The model learns to predict the label. Two flavours: regression (predict a number, e.g. house price) and classification (predict a category, e.g. spam / not spam).
• Unsupervised learning — no labels; the model finds structure, e.g. clustering customers into groups.
• Reinforcement learning — an agent learns by trial and error, receiving rewards, e.g. game-playing or robotics.

Most business ML you'll meet is supervised. Knowing which paradigm fits your problem is the first modelling decision.