StatQuest with Josh Starmer
by StatQuest / Josh Starmer
The concept you are stuck on, rebuilt from a toy dataset until it is obvious. BAM!
Overview
Josh Starmer spent thirteen years at the University of North Carolina at Chapel Hill, first as a postdoc and then as a faculty researcher, applying statistics to genetic data, and StatQuest began as an attempt to explain those methods to colleagues who were geneticists rather than statisticians. That origin explains the method: each video takes one concept, builds it from a concrete toy dataset, and refuses to introduce notation before the idea it stands for. The channel launched in May 2011 and now holds roughly 300 videos and about 1.67 million subscribers against around 93 million lifetime views. Coverage runs the whole ladder - p-values, confidence intervals, the t-test and ANOVA; linear and logistic regression; PCA, t-SNE and UMAP; decision trees, random forests, AdaBoost, gradient boost and XGBoost; support vector machines and the kernel trick; then gradient descent, backpropagation, neural networks, word embeddings, RNNs, LSTMs, attention and transformers. Longer topics are built as ordered series where each part depends on the one before, so a part one is a genuine starting point rather than a fragment dropped by the recommendation algorithm. Everything on YouTube is free; the paid extensions are three illustrated books - machine learning at 305 pages, neural networks and AI at 363 pages with PyTorch tutorials, and statistics at 319 pages - plus nine per-topic study guides and a set of Python notebooks. Starmer also teaches two applied courses on Coursera.
At a Glance
- Topic
- ML
- Level
- All Levels
- Format
- Video
- Cost
- Free
- Duration
- ~300 videos, mostly 10-30 minutes each; self-paced
- Provider
- StatQuest / Josh Starmer
- Hands-on
- No
- Certificate
- None
What You’ll Learn
- ✓The intuition behind p-values, confidence intervals, t-tests and ANOVA
- ✓How linear and logistic regression are fitted and how to read their output
- ✓How decision trees, random forests and gradient boosting actually choose splits
- ✓What PCA, t-SNE and UMAP do to high-dimensional data and where each misleads
- ✓How gradient descent and backpropagation train a neural network step by step
- ✓How word embeddings, RNNs, attention and transformers build on one another
- ✓How to read the statistical notation that textbooks assume you already know
Highlights
- •Roughly 300 videos and 1.67 million subscribers built over fifteen years - your stuck concept is almost certainly covered
- •Ordered multi-part series, so a topic can be started at part one instead of mid-thread
- •Made by a former UNC-Chapel Hill computational-biology researcher, not a general content channel
- •Every explanation is anchored to a small worked example before any formula appears
- •Free on YouTube, with optional illustrated books and study guides for the same material on paper
Who It’s For
Best For
- ✓Engineers who can run scikit-learn but cannot explain what the model did
- ✓Anyone blocked on a formula partway through a paid machine-learning course
- ✓Career switchers rebuilding statistics fundamentals before starting ML
Prerequisites
- •High-school algebra - the videos build everything else from there
- •None for the statistics series; the deep-learning series assumes the earlier ones
FAQ
What is StatQuest with Josh Starmer?
Josh Starmer's YouTube channel explains statistics and machine learning one concept at a time, always starting from a small concrete example and refusing to show notation before the idea it stands for. It covers the full ladder from p-values and regression through trees, boosting and dimensionality reduction to backpropagation, attention and transformers, in ordered multi-part series you can start at part one.
Is StatQuest with Josh Starmer free?
StatQuest with Josh Starmer is free to access.
What level is StatQuest with Josh Starmer for?
StatQuest with Josh Starmer is aimed at a all levels audience. Recommended background: High-school algebra - the videos build everything else from there, None for the statistics series; the deep-learning series assumes the earlier ones.
How long does StatQuest with Josh Starmer take?
Expect roughly ~300 videos, mostly 10-30 minutes each; self-paced. Most learners work through it at their own pace.
What will I learn from StatQuest with Josh Starmer?
You'll learn: The intuition behind p-values, confidence intervals, t-tests and ANOVA; How linear and logistic regression are fitted and how to read their output; How decision trees, random forests and gradient boosting actually choose splits; What PCA, t-SNE and UMAP do to high-dimensional data and where each misleads; How gradient descent and backpropagation train a neural network step by step; How word embeddings, RNNs, attention and transformers build on one another; How to read the statistical notation that textbooks assume you already know.
Topics
Sources
This page was written from 4 sources, 3 on domains other than youtube.com.