Why I built this site
I have spent ten years explaining machine learning in classes and meetings. This site is where I write those explanations down.
I have taught machine learning classes since 2016, and at work I spend a good part of my time explaining models to people who have to make decisions with them. In both cases the same thing happens: a formula alone convinces nobody. What works is moving a parameter and seeing what changes.
That is why the Concepts section has visualizations you can manipulate. You can change the C parameter of an SVM and see how the margin moves, prune a decision tree until it stops overfitting, or rotate a 3D point cloud before projecting it with PCA. Each concept has the intuition, just enough math, and the typical mistakes I see in class.
On the blog I will write about what does not fit there: how to build a data team from scratch, what worked and what did not when taking GenAI to production in a bank, and how to measure whether a model really moved anything in the business.
If something is poorly explained or you want me to add a concept, write to me.