Supervised machine learning uses:

Concept

In supervised learning, each training example associates input features with a known target label or numerical value. The algorithm learns a mapping by comparing its predictions with these known targets and adjusting the model to reduce prediction error.

Application

Labeled data provides both parts required for this process: the input and its known target. Therefore the model can calculate how far a prediction is from the target and learn from that error.

Contrast

• Random data describes how data may be sampled or ordered, not whether targets are attached.

• Unlabeled data contains inputs but no known target for each example.

• No data provides no observations from which to learn.

Cross-check

If the target column is removed from the training examples, the algorithm can no longer compare each prediction with a known outcome. This confirms that the distinguishing training resource is labeled data.

Result: Supervised machine learning uses labeled data.