THE CORE AIM OF VOCABULARY MASTERY · SUPERVISED LEARNING VOCABULARY · LABEL → TRAIN → PREDICT → VALIDATE → GENERALISE
Supervised learning vocabulary is the language used to describe machine-learning methods that learn from examples paired with known targets. Highly useful terms include labelled data, feature, target, classification, regression, training set, validation set, test set, loss and generalisation.
The core aim of vocabulary mastery for supervised learning vocabulary is example-to-prediction clarity. Learners should be able to explain what information enters the model, what target it is learning to predict, how prediction error is measured and whether performance continues on unseen examples.
This page is the Supervised Learning Vocabulary owner inside the eduKateSG Vocabulary hub. It connects directly to Machine Learning Vocabulary, Neural Networks Vocabulary and Deep Learning Vocabulary.
Central proposition: supervised-learning vocabulary is mastered when the learner can trace labelled examples through training to predictions on data the model has never seen.
The 60-Second Supervised Learning Vocabulary Router
- Data: example, feature, label, target.
- Tasks: classification, regression.
- Splits: training, validation, test.
- Learning: prediction, loss, optimisation.
- Evaluation: accuracy, precision, recall, error.
- Behaviour: overfitting, underfitting, generalisation.
Classification and Regression Are Different
Classification predicts categories, such as spam versus legitimate email. Regression predicts a numerical quantity, such as demand or temperature. Both can be supervised learning because known targets guide training.
Training, Validation and Test Data
The training set is used to fit model parameters. The validation set helps choose settings or compare candidate models. The test set provides a final held-out estimate of performance. Keeping these roles distinct protects evaluation from accidental leakage.
A Worked Example: Feature and Target
If a model predicts house price, floor area and location might be features, while sale price is the target. Vocabulary mastery begins by naming exactly what the model knows and what it must predict.
Overfitting and Generalisation
Overfitting means learning training examples too specifically. Generalisation means performing usefully on unseen examples drawn from relevant conditions. Good training performance alone therefore does not prove a useful model.
How to Learn Supervised Learning Vocabulary
- Label features and targets in real datasets.
- Compare classification with regression.
- Create training, validation and test splits.
- Inspect errors, not only headline accuracy.
- Compare training and validation performance.
- Explain what unseen-data performance means.
Common Mistakes
Confusing feature and label
Repair: separate information available to the model from the answer it is learning.
Testing on training data
Repair: preserve genuinely held-out examples.
Treating accuracy as universal
Repair: choose metrics that match the task and consequences.
Frequently Asked Questions
What is supervised learning?
It is machine learning from examples paired with known target outputs.
What is labelled data?
It is data where examples include the target value or category the model should learn to predict.
What is generalisation?
It is useful performance on unseen data beyond the examples used for training.
The Supervised Learning Vocabulary Standard
Mastery means identifying the features, targets, data splits, prediction task, loss and evidence of generalisation.
That is the standard: learning language precise enough to separate memorising examples from predicting new ones.