Supervised learning: numbers and classes

Choose your learning level

BeginnerIntermediateAdvanced

Start with plain-language explanations. Switch levels at any time.

Beginner explanation

Supervised learning uses examples paired with desired answers. Regression predicts a number, such as demand. Classification predicts a category, such as whether a message is spam. The model learns from labels that can themselves contain mistakes.

Intermediate explanation

Choose metrics that match the output. Regression may use MAE; classification may use precision and recall. A probability score is converted into an action using a threshold. Moving the threshold changes the balance of missed cases and false alarms.

Advanced explanation

Under class imbalance, accuracy can hide poor minority-class performance. Use confusion matrices and threshold-specific costs. Fit preprocessing only on training data, then apply it to validation and test data to avoid leaking information through the pipeline.

PUT IT IN CONTEXT

A worked example

If only 2 of 100 messages are spam, labeling everything “not spam” is 98% accurate but catches none of the spam. For spam detection, that impressive-looking accuracy is not sufficient.

TRY IT YOURSELF

Practice the idea

A detector finds 8 real spam messages, flags 2 legitimate ones and misses 4 spam messages. Calculate precision and recall.

Your working notes (not saved or sent)

Reveal the answer guide

Precision = 8/(8 + 2) = 80%. Recall = 8/(8 + 4) ≈ 66.7%. Precision asks how often a flagged item is truly spam; recall asks how much of the actual spam was found.

Go further: advanced challenge

Choose a threshold policy for a review queue with limited staff. Explain how review capacity changes the desirable precision–recall tradeoff.

An open-ended discussion task. Explain your assumptions and evidence; there may be more than one defensible answer.

Continue the conversation

What part of this idea would you like to test or question? Compare your reasoning with the community.

Find related forum discussions ↗

Opens a topic search in the existing AstraCruz forum. Results depend on available posts; signing in may be required. No discussion or activity is fabricated.

Browse all forums