The maths ML actually uses
Enough linear algebra, calculus and probability to read ML courses without panic
Plan it around your week
This path is roughly 24 hours in total. Tell us how much time you actually have.
About 6 weeks.
Based on our own estimate of the path, not the providers' figures. Most people need a bit longer than they think.
How long does it take?
| Courses | 2 |
|---|---|
| Weeks | About 6 |
| Hours a week | 4 |
| Total hours | About 24 |
The steps, in order
Why this step: Introduces the maths visually before you need to derive anything by hand.
You're ready for the next step when: You recognise gradients, loss surfaces and matrices in context.
Why this step: Applies probability and optimisation to real algorithms so the maths has a purpose.
You're ready for the next step when: You can follow the maths in most intermediate ML courses.
Where to go next
- All free AI learning paths
Every path, with who it suits and how long it takes.
- All free AI courses, rated
Filter the whole catalogue by level, topic and certificate.
- AI learning path for non-coders
Explain how modern AI works and use it well at work
- Use AI tools well at work
Prompt, evaluate output, and automate everyday tasks safely
- Python for AI, from zero
Write the Python every AI course assumes
- Machine learning foundations
Train, evaluate and explain classic ML models
- Deep learning, hands-on
Build and train neural networks for vision and text
- Build apps with LLMs
Ship an app that uses an LLM API, retrieval and evaluation
- Become an ML engineer
Deploy, monitor and maintain models in production
- From data analysis to AI
Move from spreadsheets and SQL to ML and AI
- Is Google worth it?
Our review of Google, where steps on this path are hosted.
- Is Harvard CS50 worth it?
Our review of Harvard CS50, where steps on this path are hosted.
Want the bigger picture?
Roadmaps show every skill in a field, in order, so you can see where this path fits.