Skip to content
TechRank home

Glossary

What is Machine Learning?

A way of building software that learns patterns from examples instead of following rules written by hand, used for prediction, classification and recommendation.

In plain words

Ordinary software follows rules a programmer wrote. Machine learning flips this: you show a program many examples, with the right answers, and it works out the rules itself. Show it thousands of past loan applications and what happened to each, and it learns what separates the ones that were repaid from the ones that were not.

Machine learning is the wider field. Language models, image recognition and recommendation engines are all kinds of it.

The main kinds

  • Supervised learning: learns from examples that include the correct answer, for example spam and not spam
  • Unsupervised learning: finds groups or unusual items in data with no answers given
  • Reinforcement learning: learns by trying actions and being rewarded, as in games and some control systems

Common uses

  • Predicting demand, churn or equipment failure
  • Flagging suspicious payments
  • Recommending products or content
  • Scoring and sorting leads
  • Reading images and text (see computer vision and NLP)

When you do not need it

If a clear rule solves the problem, such as "orders over a set amount need approval", write the rule. Machine learning earns its place when the pattern is too complicated to write down and you have enough good examples of past outcomes. Too little data, or data that does not reflect the future, is the most common reason projects fail.

Common mistakes

  • Starting with the technique instead of the business question
  • Not enough data, or data full of errors and gaps
  • Judging success on the data the model was trained on instead of fresh examples
  • No plan for keeping it accurate after launch (see MLOps)

Questions to ask a company that offers this

  1. What exactly will the model predict, and how will we know it is good enough?
  2. How much data do we need, and what condition does it have to be in?
  3. How will you test it on examples it has never seen?
  4. Would a simple rule or report do the same job?
  5. Who maintains it after launch?

A good company will answer these plainly and in writing. Vague or evasive answers are worth noting.

Related terms

Find companies

Explanation last read through on 13 October 2026. It is general guidance in plain language, not advice for your particular project. Tell us if something is wrong or unclear.