Machine Learning from Scratch

Classical machine learning in Python, from numpy and pandas to models you can train, evaluate, tune and explain.

Basic19 modules217 lessons38 exercises≈ 15 h

Tests check every exercise: in your browser and in the cloud

Exercises run in the browser, so you need nothing installed to take the course. To run lesson code and exercises on your own computer, set it up in four steps.

Install Python 3.12
You need exactly 3.12: the library versions the site runs have ready-made builds for it. Download the installer for your system from python.org.
Install VS Code
Download VS Code and add the Python extension from Microsoft.
Create an environment and install the libraries
Open a folder for the course in VS Code and create a virtual environment in it. Install the versions the site runs, so your output matches the lessons: numpy 1.26, pandas 2.2, scikit-learn 1.4 and pytest.
Run lesson code and exercises
Save a lesson's code to a file and run it with the ▶ button. For an exercise, save both editor tabs under their names and run the tests from the Testing panel.

Set up your computer

Install Python 3.12

Install VS Code

Create an environment and install the libraries

Run lesson code and exercises

Syllabus

01What Machine Learning IsFree≈ 15 min9 lessons

02NumPy and pandasFree≈ 50 min12 lessons

03Preparing Data≈ 50 min12 lessons

04Linear Regression≈ 1 h 5 min15 lessons

05Beyond a Straight Line≈ 50 min12 lessons

06Logistic Regression≈ 50 min13 lessons

07Neighbors and Naive Bayes≈ 50 min12 lessons

08Support Vector Machines≈ 45 min10 lessons

09Trees and Forests≈ 50 min12 lessons

10Choosing and Tuning Models≈ 50 min12 lessons

11Gradient Boosting≈ 50 min12 lessons

12Clustering≈ 45 min11 lessons

13Dimensionality Reduction≈ 50 min12 lessons

14Association Rules≈ 40 min9 lessons

15Bandits and Smarter A/B Tests≈ 50 min12 lessons

16Classic Text Processing≈ 50 min12 lessons

17Neural Networks≈ 50 min12 lessons

18Shipping Models≈ 50 min12 lessons

19Final Project≈ 50 min6 lessons