Machine learning means teaching computers to find patterns in data instead of writing every rule by hand. Two of its main approaches are supervised and unsupervised learning.

Supervised learning

In supervised learning, the training data comes with answers, called labels. A system might be shown many emails marked as spam or not spam, and learn to predict the label for new emails. Predicting house prices from past sales is another example. The model learns the link between inputs and known outcomes.

Unsupervised learning

In unsupervised learning, there are no labels. The system looks for structure on its own, such as grouping customers with similar buying habits or spotting unusual transactions. Humans then interpret what the groups or patterns mean.

Choosing between them

  • Use supervised learning when you have reliable labelled examples and a clear prediction goal.
  • Use unsupervised learning to explore data or discover groups when you do not know the answers in advance.
  • Many real projects combine both approaches.

A word on other approaches

Reinforcement learning, where a system learns by trial and reward, and semi-supervised learning, which mixes small amounts of labelled data with larger unlabelled sets, are also widely studied.

Takeaway

The right method depends on the question you are asking and the data you have. Understanding the difference helps non-experts judge claims about what a machine learning system can really do.