StatQuest with Josh Starmer

What StatQuest is

StatQuest is Josh Starmer's YouTube channel and website, described on statquest.org as "an epic journey through data science, statistics, machine learning, neural networks, and AI". Each video takes one idea and builds it up with simple drawings and small worked examples, so you see what a method actually does to the data before you meet the notation.

The official video index groups the videos into clear sections:

• Statistics fundamentals: histograms, distributions, the normal distribution, hypothesis testing, the null hypothesis, p-values and correlation.
• Statistical tests: significance testing, Fisher's exact test and the Benjamini-Hochberg procedure for multiple comparisons.
• Machine learning: PCA step by step, clustering, decision trees, random forests, XGBoost and support vector machines.
• Neural networks, deep learning and AI: from the basics through convolutional networks to the transformers behind large language models.

Is it worth your time?

Yes, as a companion rather than a full course. StatQuest is strongest at the moment you are stuck: a textbook mentions the bootstrap, ridge regression or a false discovery rate, and you want to understand the idea in fifteen minutes. The videos favour intuition over proofs, which is exactly what most learners lack at first. Pair them with a book that gives you exercises and code, and you get both the picture and the practice.

Where it fits for a data science learner

Use it alongside your statistics and modelling reading. Watch the statistics fundamentals before you run your first hypothesis test, the regression and regularisation videos while you work through a statistical learning textbook, and the tree and boosting videos before you tune a model you do not yet understand. The multiple testing videos are also a useful warm-up before you analyse A/B experiments.

How much time does it take?

Most videos are short enough to watch in one sitting, and many topics are split into parts. There is no fixed order to finish: pick the section you need, watch it once for the idea, then again with a notebook open and try to reproduce the example by hand or in code.

Free vs paid

The videos are free on YouTube and linked from the video index. StatQuest also sells optional study guides through Gumroad, and Josh Starmer has written illustrated books, including The StatQuest Illustrated Guide to Neural Networks and AI with hands-on PyTorch tutorials. None of these are required to learn from the videos.

Prerequisites

School level algebra is enough for the statistics videos. The machine learning and neural network sections assume you have seen the earlier basics, and the index links back to them when a video depends on another.

Common questions

Can StatQuest replace a textbook? Not on its own. It gives you intuition, not practice problems or a complete syllabus. Use it next to a book or course that makes you write code and answer questions.

Is it only for beginners? No. Experienced practitioners use it to refresh a method quickly or to find a clear way to explain one to a colleague.

Best for: anyone who has read a definition three times and still cannot say what the method is doing.