Skip to content
BENIBIX Learn

Classification (Machine Learning)

Also known as Classifier, Classification Task

Sorting each case into one of a few known groups, like spam or not spam.

Think of it like

Think of it like sorting socks into the right drawers. Each sock goes in one drawer.

Example

Jack's model looks at photos of fruit and says apple, banana or pear. Each photo gets exactly one of those three answers.

Why it matters

Many real jobs are a choice between known options. Spam, fraud, illness and faults are all classification.

Definition

Classification is the machine learning task of putting each example into one of a set of groups. The groups are fixed before training, like spam or not spam, or cat, dog or bird. The model learns from labelled examples of each group. For a new example, it picks the group that fits best. It often gives a chance for each group too.
Classification is the left panel. The answer is one of a few named groups.

Common mistake

Judging it only by how often it is right. If 99 in 100 emails are fine, a model that says fine to all scores 99 percent and catches nothing.

Key takeaways

  1. 1
    Known groups. Set before training.
  2. 2
    One answer each. Pick the best fit.
  3. 3
    Often a chance. How sure it is for each group.

Compare

Learn it properly