Supervised vs Unsupervised; Train/Test

Machine learning is the art of letting a program learn a pattern from data instead of being told the rule by hand. Before any formula, you only need one distinction: does your data come with answers attached, or not? That single question splits the entire field in two.

Supervised vs unsupervised — the great divide

In supervised learning each training example carries a known answer, called a label. The model sees inputs paired with the correct output and learns to reproduce that mapping on new, unseen inputs. If the label is a continuous number (a price, a temperature) the task is regression; if it is a discrete class (spam or not-spam, the digit 0-9) it is classification.

In unsupervised learning there are no labels — just the inputs. The model hunts for structure on its own: clustering groups similar points together, and dimensionality reduction compresses many features into a few while keeping the important variation.

Features, labels, and what a model learns

A dataset is a table. Each row is one example; each column is a measured quantity. The input columns are the features (often written X), and in supervised learning the answer column is the label or target (y). A supervised model learns a function that maps features to label, f(X) → y, so that when a brand-new row arrives with features but no answer, it can predict one. Unsupervised methods see only X and must invent the structure themselves.

Train, validation, test — and the golden rule

You cannot judge a model by how well it memorizes the data it studied — anyone can ace an open-book test on questions they have already seen. So we split the data into three disjoint parts:

• Training set — the model fits its parameters here.
• Validation set — used to tune knobs (which model, how complex) during development, so you can compare options without touching the test set.
• Test set — a sealed sample, opened once, to estimate how the model will behave on genuinely new data. This number is your honest generalization score.

A model that does great on training data but poorly on unseen data is overfitting — it memorized noise instead of the signal. The opposite, underfitting, is a model too simple to capture the real pattern, so it does poorly even on the training data. The validation set is how you steer between the two.

How GATE asks this

This topic is almost always a conceptual MCQ or MSQ: classify a list of tasks as regression / classification / clustering, or pick the true statements about the train-test split (for example, “the test set may be used to tune hyperparameters” is false). Occasionally a tiny NAT asks for a split size — if 1000 samples are split 80/20, the test set has 200. Read every option for the word that fixes the family: a continuous target means regression, a class label means classification, no labels at all means unsupervised.

Worked example — name that task

Classify each task by family:

House-price prediction from size & location  →  supervised · REGRESSION
        (target is a continuous number — a price)

Email spam detection (spam / not-spam)        →  supervised · CLASSIFICATION
        (target is a discrete label — two classes)

Customer segmentation from purchase history   →  unsupervised · CLUSTERING
        (no labels given — the groups are discovered)

Compressing 100 sensor readings to 2 axes     →  unsupervised · DIM. REDUCTION
        (no target — just summarise the variation)

The deciding question each time: is there a known answer column? If yes, is it a number (regression) or a category (classification)? If no, the task is unsupervised.

Quick check