Regression is used when the thing you predict is a continuous number.
Examples are house price, temperature, revenue or demand.
The model learns a relationship between input features and a numeric target.
Predicting salary from years of experience is a classic small example.
Linear regression is the simplest algorithm in this family.
Regression means the answer is a number on a scale.
Price, temperature, or how many units will sell next month.
The model can be wrong by a little or by a lot, and the size of the gap matters.
That is the main difference from predicting a category.
Being wrong by two pounds is not the same as being wrong by two hundred.
So every error measure for regression is really about distance.
You subtract the guess from the truth and then do something with the gap.
Predictions are not limited to values that appeared in training.
A model will happily predict a negative price if nothing stops it.
That usually means your features do not cover the situation you asked about.
Regression gets weak at the edges of the data it has seen.
Predicting house prices far above anything in your training set is guesswork.
Always plot predictions against the truth before trusting a single number.
A good average error can hide a model that fails badly on expensive houses.
If the thing you predict is heavily skewed, try predicting its log instead.
One question on this card. Every option explains itself once you pick, right or wrong.
55 concepts in this module, then a graded test with every answer explained.