Supervised learning Introduction
Linear regression
Predict a numerical target with a weighted sum.
Before this
The intuition
Linear regression fits weights and an intercept to predict a number from input features. Fitting by least squares minimizes the sum of squared residuals.
A concrete example
A toy rule price = 2 × area + 10 predicts 170 for area 80. If the observed target is 180, the residual target − prediction is 10.
What to remember
A fitted relationship does not demonstrate causation. Predictions far outside the training range can be unreliable.
Check your understanding
What does the intercept predict when every feature is zero?
Reveal explanation
The intercept itself, though zero may not be a meaningful real-world input.