Classification and Regression

Supervised learning splits in two, and the split is about what comes out. A word from a fixed list, or a number that could be anything.

1The question to ask

Look at the answer, not the question

If the output is one of a fixed set of named groups — hot or cold, spam or not spam — it is classification. Classification works on discrete data.
If the output is a number that could land anywhere in a range — a temperature, a price — it is regression. Regression works on continuous data.

Both of these live under Supervised

Switch between Classification and Regression. Notice they share a parent — both need labelled data.

Supervised. The answer is a category: discrete data.

Full lineage: AI Models → Learning Based → Machine Learning → Supervised → Classification

2Classification

The data is sorted according to its labels. Think of the grading system at school: students are put into grades on the basis of the marks they got. The output is one of a fixed set of grades — the data need not be continuous.

Hot or cold tomorrow?

The model is trained on past weather data in which each day was labelled “hot” or “cold”. It learns which patterns separate the two, using things like location, season, the high and low temperatures and the humidity. Given tomorrow’s figures, it puts tomorrow in one of those two boxes.

Spam or not spam?

The model is shown a great many emails, real ones and spam. It learns what makes an email look like spam. Shown a new one, it assigns a category — exactly like sorting post into two piles. Modern email has grown more boxes to sort into: social, advertisements, notifications.

3Regression

Regression models work on continuous data and predict a continuous value from the input variables — temperature, price, income, age.

What will this house sell for?

The features (the independent variables) are the number of bedrooms, the carpet size and the garage area. The label (the dependent variable) is the price. The model learns the link and can then price a house it has never seen.

What is this used car worth?

Fuel type, years of service, number of previous owners, kilometres driven, manual or automatic. The model predicts an approximate price — a continuous value — so it is a regression model.

Pigeonholes and a tape measure

Classification hands you a pigeonhole: this letter goes in that slot, and there are only so many slots. Regression hands you a tape measure: the answer is 1.63 metres, and it could just as easily have been 1.64.

4Side by side

	Classification	Regression
Output	A category from a fixed set	A value from a continuous range
Kind of data	Discrete — it need not be continuous	Continuous
Target variable	Categorical	Numerical
Example	Is this email spam?	What will tomorrow's temperature be?
Example	Is this applicant eligible for the loan?	What is this house worth?

Careful — the same subject, two different models

“Will tomorrow be hot or cold?” is classification: two named boxes.
“What will tomorrow’s temperature be?” is regression: a number.
Same weather, same data. The output is what decides, and that is precisely the trap.

5From the board papers

Which scenario best represents a regression problem?
(A) Identifying whether an email is spam. (B) Grouping customers by behaviour. (C) Predicting tomorrow’s temperature. (D) Recognising faces in photos.CBSE board exam 2026

1 mark

▶Show the answer

(C) Predicting tomorrow’s temperature. A temperature is a continuous value. (A) and (D) put things into categories, so they are classification; (B) forms groups nobody defined, so it is clustering.

Give any two characteristics of a Classification Model.CBSE board exam 2026

2 marks

▶Show the answer

1. It is a supervised learning model: it is trained on labelled data, and it sorts the data according to those labels.
2. Its output is a category from a fixed set — it works on discrete data, so the data need not be continuous.

A healthcare provider has a dataset of patient demographics, medical history and treatment details. Name the type of model for each task:
(a) predict whether a patient will have a short or long hospital stay; (b) predict the number of days a patient will stay.Class X handbook, Reflection Time

2 marks

▶Show the answer

(a) Classification. “Short” and “long” are two named categories.
(b) Regression. A number of days is a continuous value.

6Check yourself

Quick Check

The target variable is categorical in which kind of problem?

Quick Check

Which model would you use to predict a continuous valued output?

Quick Check

Predicting whether a customer is eligible for a bank loan or not is:

Quick Check

A real estate agency predicts house selling prices from size, location and number of bedrooms. This is:

← Previous

🎓 Supervised, Unsupervised, Reinforcement

Clustering and Association 🫧