R2D3 visual ML tutorial explains decision trees

A decade on, this remains the gold standard for explaining ML intuitively — and the fact it's still circulating on Hacker News says something about how poorly most ML education has aged.

• –Uses SF vs. NYC home classification to build intuition for decision trees, splits, training/test sets, and overfitting — entirely through scroll-triggered animations
• –Zero equations: the entire explainer runs on visual metaphors and concrete data, making it accessible to designers and non-engineers entering AI work
• –Shortlisted at the 2015 Kantar Information is Beautiful Awards and garnered 250k visitors in its first month
• –Part 2 extends the series into model bias and the bias-variance tradeoff
• –Tony Chu (co-creator) now works at Augment Code, an AI coding startup — a fitting arc for someone who helped define visual ML pedagogy