Learning path
Next moduleModule 4: Classification
Module 3
Regression
Predict continuous values with linear models, regularize them, and choose the right error metric.
- 1. Linear regressionFit a line (or plane) to data by minimizing squared error, and compute one gradient-descent step by hand.40 min
- 2. Regularization and regression metricsControl linear models with ridge, lasso, and elastic net, compute their effect on a weight by hand, and choose between MAE, RMSE, R², and MAPE.35 min
Regression predicts a number: a price, a demand forecast, a latency. This module builds the simplest useful model, linear regression, all the way from intuition to a gradient-descent update you compute by hand. It then shows how penalties (ridge, lasso, elastic net) keep weights under control and how to choose between MAE, RMSE, R², and MAPE.
| # | Lesson | You will be able to |
|---|---|---|
| 1 | Linear regression | compute predictions, residuals, MSE, gradients, and one update step by hand |
| 2 | Regularization and regression metrics | explain how L1 and L2 change weights, and pick a metric that matches error cost |
Interactive lab: Linear regression: drag points, see residuals, toggle L1/L2.
Code: code/03-regression/