Core concepts of learning from data.
Learning from labeled data for regression and classification.
Finding hidden patterns in unlabeled data, such as clustering.
The fundamental tension between underfitting and overfitting.
How models measure error and learn by minimizing it.