Lesson 3 · Machine Learning
Linear Regression
8 min
You'll be able to
- Understand the equation of a line in data terms
- Explain what the model learns
- Interpret coefficients meaningfully
Linear regression models a target as a weighted sum of features plus a bias. If y is price and x is square footage, we predict y = w·x + b, where w is a weight and b a bias.
Training finds the w and b that minimise the difference between predictions and actual values — usually measured by mean squared error.
import numpy as np
# data: square footage -> price
x = np.array([800, 1000, 1200, 1500])
y = np.array([120, 150, 180, 230])
# closed-form least squares
w, b = np.polyfit(x, y, 1)
print(f"price = {w:.2f} * sqft + {b:.2f}")Challenge
Fit by hand
Given points (1,2), (2,4), (3,6), guess a simple w and b that fit perfectly. Confirm the pattern.
Knowledge Check
Linear regression
Linear regression is a type of classification task.
In y = w·x + b, what does the model learn during training?
Answer all questions to submit.