Skip to content
BENIBIX Learn

Label (Machine Learning)

Also known as Target, Target Variable, Ground Truth Label

The right answer that comes with an example. The model learns to guess it.

Think of it like

Think of it like the answer key at the back of a book.

Example

A hospital pays doctors to look at 20,000 scans. They mark each one healthy or not healthy. Those marks are the labels.

Why it matters

Labels teach the model what right looks like. Wrong labels teach it the wrong thing, no matter how good the method is.

Definition

A label is the right answer that comes with a training example. For a photo of a cat, the label is cat. For a house, the label might be the price it sold for. The model learns to guess the label from the features. Labels are often added by people, which takes time and money.
The price column is the label. It is the answer the model learns to guess from the other columns.

Common mistake

Trusting labels without checking them. Tired or rushed labellers make mistakes, and the model learns those mistakes too.

Key takeaways

  1. 1
    The right answer. What the model should say.
  2. 2
    Often made by people. Labelling costs time and money.
  3. 3
    Must be right. Bad labels teach bad lessons.

Learn it properly