Machine Learning (ML) · Beginner level

Beginner level · Module 1: Machine Learning Foundations · Lesson 1 of 3

What Machine Learning Is

See how learning from examples differs from writing rules, and meet the main types of machine learning.

In this lesson you will

  • Distinguish rule-based programs from learned models
  • Classify problems as supervised (regression or classification) or unsupervised

Key concepts

modelfeatures and labelssupervised vs unsupervisedregression vs classification

The lesson

A traditional program follows rules a person wrote. Machine learning finds the rules from examples: shown many past loan applications with their outcomes, a model learns a pattern that links the applicant's details (features) to the outcome (the label).

Supervised learning uses labelled examples. If the label is a number, such as next month's sales, it is regression; if it is a category, such as 'repaid' or 'defaulted', it is classification. Unsupervised learning finds structure in unlabelled data, for example grouping customers with similar buying habits.

ML is useful when patterns are too complex to write as rules and enough good examples exist. It is a poor choice when the rule is simple, when data is scarce or unrepresentative, or when every decision must be fully explained by law or policy.

Step by step

  1. List three decisions made in your community that use rules.
  2. Identify features and a label for one of them.
  3. Classify ten example problems as regression, classification or clustering.
  4. Name one problem where ML would be a bad idea and say why.

Practice activity

Complete the twelve-problem classification exercise from the practical activity.

Expected result: Each problem is correctly classified, with a short justification and at least one 'not an ML problem'.

Check your understanding

  1. Predicting next month's maize price is which type of problem?

    Show answer

    Regression — the label is a number.

  2. What is a label in supervised learning?

    Show answer

    The outcome the model should learn to predict.

Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.