The algorithms that ran the world before deep learning, and still run most of it. Build nearest neighbours, grow a decision tree, and cluster without labels.
3 units, about 3 hours. Free.
Implement kNN, explain the role of k, and identify when distance stops meaning anything.
2Grow a decision stump by information gain and explain entropy, overfitting and interpretability.
3Implement the k-means loop and explain initialisation sensitivity and the problem of choosing k.