The best free R stats site I’ve found. Clear explanations, live examples, and zero clutter. Absolute gold for learners.
Knowing which button to press is not enough
The only way to really understand a method is to run it, break it, and watch what it does to real data
Pick a learning track
Seven roadmaps and 368 lessons so far, with more being published. The first section of every track is free.
- 01Regression, done properly
- 02Trees and gradient boosting
- 03Feature engineering and selection
- 04Calibrated and conformal prediction
- 05Causal inference for decisions
- 06Survival and time-to-event
- 07Robustness, drift and distribution shift
- 08LLMs and modern ML
Meet the hint ladder, a mentor that asks instead of tells
Every graded exercise comes with hints that nudge you toward the idea before you ever see the answer. Reply to any daily lesson and Akshay answers in person.
mean() should do when one value is unknown: guess, or refuse? R refuses. Which argument tells it to drop the gaps before averaging?
AJoin learners from 40 universities and research institutes
600 learners have signed up since launch: people who need the statistics to be right, at work and in print
This site helped me pass my technical interview. Practical R examples gave me the confidence to use statistics at work.
Every tutorial includes reproducible code and real datasets. r-statistics taught me regression better than my grad course.
People who understand the statistics earn well over $100,000 a year in the US
Estimated total pay in the United States, base plus bonus and equity, across all experience levels. Glassdoor, 2026.
These figures include entry-level roles; the top quarter of data scientists clear $200,000 and credentialed actuaries routinely earn far more. The Bureau of Labor Statistics projects data scientist employment to grow 35 percent between 2025 and 2035, among the fastest of any occupation.
Some people look at the job data and assume analytics means spreadsheets and dashboards. On the contrary, the demand is for people who can say what a number means, how sure we are, and what would change their mind. Most of these roles do not require a specific degree, but you do need to know your stuff.
The library
Over 1,800 free tutorials, exercises, handbooks and tools, all runnable, all yours whether or not you ever pay
How do I learn statistics and ML?
Pick one tool and spend most of your time running methods on real data, not watching videos. Follow a structured path, build real projects along the way, and practice daily.
- Pick one tool and stick with it.R is a great first choice for this: the statistics are first-class, and every method you will ever read about ships with an implementation.
- Follow a structured path.Random tutorials leave gaps. A linear roadmap removes the "what should I learn next?" problem entirely.
- Run and learn something every day.Understanding is a skill that takes real time. Daily practice beats occasional binge-sessions.
- Work on real data.Real data has gaps, outliers and surprises. Projects force you to handle them, and they become the portfolio that gets you interviews.
- Get unstuck fast.Everyone gets stuck. Lean on a hint ladder that asks questions, and a human who answers when you reply.
- Go deep on one specialty.Once you have the foundations, specialise. Bayesian methods, causal inference, time series, machine learning. Pick one and go deep.
Frequently asked questions
Got questions? We've got answers
Yes. It is free to create an account and start learning. The first section of every track is free, with graded exercises, and so are all 1,800 tutorials, the handbooks and the tools.
Run the methods on real data, in order, every day. The tracks here are that order. Each page teaches one idea, runs the code in front of you, and then checks that you can do it yourself.
Yes, and most people here do. What you need is a structured path so you never wonder what comes next, feedback when you are wrong, and someone to ask when you are stuck. All three are built in.
About three months per track at three hours a week. The daily email lesson keeps the thread between sessions.
Yes. The first section of every track, the whole tutorial library, the handbooks and the tools stay free. Pro unlocks the rest of every track, including each new lesson as it is published, and every certificate.
AI can write the code. It cannot tell you whether the answer is right, whether the test fits the question, or whether the model is lying to you. That judgement is the job, and it is what these tracks teach.
An online learning platform for statistics and machine learning, taught through R. Interactive lessons, graded practice, and certificates, with a free library that has been around since 2015.
Every method is explained from the ground up and run live in your browser, with the statistics underneath the model rather than just the buttons to press. Progress is graded, not watched.
See how it feels in two minutes.
Two real lessons from a learning track, graded, in your browser. No account, nothing to install.
It takes 2 minutes. No account, nothing to install.