Machine Learning Crash Course review
Google · google
Our Google Machine Learning Crash Course review: it is the best free machine learning course for people who can already program. Google teaches 12 modules in about 13.5 hours with animated videos, interactive visualisations and Colab exercises. Nothing is behind a paywall, but you earn quiz badges, not a certificate.
What does it take?
| Difficulty | 2 out of 5 |
|---|---|
| Hours | About 13.5 hours |
| Format | Video, Interactive exercises, Code notebooks |
| Maths needed | School-level maths |
| Coding needed | Yes, you will write code |
| Hardware needed | A browser is enough |
| Certificate | No certificate |
| Free to audit | Yes |
| Pace | Self-paced |
| Last updated by the provider | Not checked yet |
| Provider | |
| Time to complete | About 13.5 hours by Google's module timings, not counting prework |
| Modules | 12, of 30 to 110 minutes each |
| Prework | Introduction to Machine Learning, NumPy UltraQuick Tutorial, pandas UltraQuick Tutorial |
| Credential | A badge for each module quiz, no certificate |
| Exercises | Colab in the browser, no setup, using NumPy, pandas and Keras |
| Running since | 2018 |
Who should take it, and who should skip it?
Take it if you are
- Programmers, ideally in Python, who want the core machine learning ideas to stick
- Learners who finished Kaggle Intro to Machine Learning and want more depth
- Anyone who wants a module on large language models alongside the classic topics
Skip it if you are
- Anyone who needs a certificate: Google gives quiz badges, not a certificate
- Anyone who has not programmed yet: take Kaggle Learn: Python first
- Learners who want machine learning APIs taught in depth: Google says the course does not do that
What will you be able to do after it?
- Explain linear regression and logistic regression, and how classification works
- Prepare numerical and categorical data for a model
- Spot overfitting and explain why datasets and generalisation matter
- Describe neural networks, embeddings and large language models
- Weigh production systems, automated machine learning and fairness
What's in each module?
| Module | What it covers |
|---|---|
| Linear regression | Google lists this module at 80 minutes. |
| Logistic regression | Google lists this module at 35 minutes. |
| Classification | Google lists this module at 70 minutes. |
| Working with numerical data | Google lists this module at 85 minutes. |
| Working with categorical data | Google lists this module at 50 minutes. |
| Datasets, generalization, and overfitting | Google lists this module at 105 minutes, the second longest. |
| Neural networks | Google lists this module at 75 minutes. |
| Embeddings | Google lists this module at 45 minutes. |
| Large language models | Google lists this module at 45 minutes. |
| Production ML systems | Google lists this module at 80 minutes. |
| Automated machine learning | Google lists this module at 30 minutes, the shortest. |
| Fairness | Google lists this module at 110 minutes, the longest. |
What we liked, and what we didn't
What we liked
- Nothing is behind a paywall: all the pages are open on developers.google.com
- Animated videos, interactive visualisations and practice exercises in one course
- Exercises run in Colab in your browser with no setup
- The modules are self-contained, with Google's own minutes shown for each
- Google has run it since 2018, and the refreshed version covers recent advances, including a large language models module
What we didn't
- There is no certificate: each module's quiz earns a badge and that is all
- It is not a first course: Google lists prerequisites and says you should be a good programmer
- The prework is extra: Google's Introduction to Machine Learning course and the NumPy and pandas UltraQuick tutorials
- Google says it does not teach machine learning APIs in depth
- Some modules are long: Fairness is 110 minutes and Datasets, generalization, and overfitting is 105
What's the catch?
There is no certificate: each module's quiz earns a badge, and that is the only credential. It is also not a first course. Google lists prerequisites and prework (its Introduction to Machine Learning course, a NumPy tutorial and a pandas tutorial) that are not in the 13.5 hours. If that is too much, take Kaggle Intro to Machine Learning first: it is 3 hours and gets you to a working model.
Is it still current?
Google describes a refreshed version that covers recent advances in AI, and it now includes a 45 minute Large language models module. The course has run since 2018, and Google's pages were checked on 4 October 2026.
What should you take before it?
- Kaggle Intro to Machine Learning
Kaggle's 7-lesson first machine learning course takes about 3 hours, costs nothing and ends with a competition. Kaggle gives a free completion certificate.
What should you take after it?
- CS50's Introduction to AI with Python
CS50's Introduction to AI with Python runs 7 weeks with 12 projects, costs nothing and gives a free CS50 Certificate if you score 70% on every project.
- AI for Beginners
A 24-lesson open curriculum on GitHub covering classic AI, ML and a little deep learning.
How we rated it: 8.5 out of 10
| Criterion | Score out of 10 |
|---|---|
| Clarity | 8 |
| Hands-on practice | 8 |
| Up to date | 8 |
| Depth | 7 |
| Truly free | 10 |
Questions people ask
Is Google's Machine Learning Crash Course free?
Yes. All the pages are open on developers.google.com, and the exercises run in Colab in your browser with no setup.
Does Machine Learning Crash Course give a certificate?
No. Each module's quiz earns a badge, for example the Linear Regression badge, but Google does not offer a certificate.
How long is Machine Learning Crash Course?
About 13.5 hours by Google's own module timings, which add up to 810 minutes across 12 modules. That does not count the prework, so plan for more.
What do I need before Machine Learning Crash Course?
Google asks you to be comfortable with variables, linear equations, graphs of functions, histograms and statistical means. It says you should be a good programmer, ideally in Python. Calculus is optional. If that sounds like a stretch, Kaggle Learn: Python and Kaggle Intro to Machine Learning are the gentler route in.
Do I have to take the modules in order?
No, each module is self-contained. Google recommends that anyone new to machine learning completes them in the order listed: from Linear regression through to Fairness. If you already know the basics, you can jump to a single module, such as Large language models at 45 minutes.
What is the prework for Machine Learning Crash Course?
Google lists its Introduction to Machine Learning course, the NumPy UltraQuick Tutorial and the pandas UltraQuick Tutorial. The exercises use NumPy, pandas and Keras, and they run in Colab in your browser with no setup.
So should you take it?
Take Machine Learning Crash Course if you can already program and want the ideas explained properly. It scores 8.5: hands-on is 8 because the exercises run in Colab, depth is 7 because 12 modules run from linear regression to fairness, and truly free is 10 because every page is open. It is not a first course: Google expects you to be comfortable with variables, linear equations, graphs of functions, histograms and means, and says you should be a good programmer.
Where to go next
- Machine learning foundations
A learning path that includes Machine Learning Crash Course.
- The maths ML actually uses
A learning path that includes Machine Learning Crash Course.
- From data analysis to AI
A learning path that includes Machine Learning Crash Course.
- Google: our platform review
What else it offers, and what costs money.
- Generative AI for Everyone
What large language models can actually do for you and your team, without writing a line of code. Andrew Ng teaches it in 6 hours on Coursera, certificate paid.
Last checked : module list and timings, prerequisites, prework, tools, badges and free access.
Not sure this is the one?
Answer two questions and we'll point you to the course that fits where you are today.