Machine Learning, Foundations

Nine lessons on how a machine actually learns from data: what fitting means mechanically, why the features you choose matter more than the algorithm you pick, how to hold out data honestly, why accuracy lies on rare events, what the classic algorithms are good at, and how gradient descent and neural networks follow from all of it. You should be able to read code. You do not need calculus, statistics, or a maths degree — every idea arrives with real numbers attached.

1. 1What a machine actually learns8 minLearning means adjusting the parameters of a shape you chose until a chosen error number is as small as it can be.
2. 2With answers, and without7 minA supervised model learns your labels, not the reality behind them, so who made the labels is part of the model.
3. 3Features beat models9 minAsk of every feature: at the moment I need the prediction, does this value already exist with this value?
4. 4Train, validation, test, and the discipline of not looking9 minThe test set is a single-use measuring device; every peek turns it into another validation set.
5. 5Overfitting, underfitting, and how to tell8 minCompare training error with validation error; the gap between them, not the score itself, names the problem.
6. 6When accuracy lies9 minChoose the metric that matches what each kind of mistake costs, then set the threshold yourself.
7. 7The classic algorithms, and when each is right10 minOn tabular data, use linear models as a floor and gradient boosting as the answer; everything else needs a stated reason.
8. 8Gradient descent, or walking downhill in fog8 minGradient descent only knows the slope where it is standing; the step size decides whether that knowledge helps or destroys it.
9. 9Neural networks as the natural next step9 minA neural network learns its own features; the loss, the splits, the metrics and the thresholds do not change.