Phase 3 · Machine learning · Models
Linear regression
Fit a line, and predict numbers from features.
Open this lesson Create a free account
Why it matters
Linear regression is the workhorse of prediction and the clearest model to learn from. It predicts a number as a weighted sum of features, so each weight is directly readable: "each extra order adds about $X of lifetime value, holding the rest fixed". It's also exactly what one neuron computes before its activation, so everything here carries straight into neural networks.
What you'll be able to do
- Write the linear regression model and read its coefficients
- Fit it by least squares, by hand and with scikit-learn
- Measure fit with R² and look at residuals
- Know its limits: straight-line assumptions, outliers and extrapolation
Before this lesson
Next in this phase: Loss functions