Animated · narrated · interactive
Each chapter is one animated picture with a voice walking you through it, quick checks inside the drawing, and practice at the end.
Regression, classification, resampling, trees, support vector machines, deep learning, survival analysis and unsupervised methods, explained from the pictures up.
The deeper treatment: linear methods, basis expansions, kernel smoothing, model assessment, boosting, neural networks and beyond.
The linear algebra, geometry, calculus and probability that the learning methods are built on.
Neural networks from the linear layer to convolutions, recurrence, attention and optimization, plus the maths behind them. Ideas rather than code.