A working introduction to applied machine learning for people who will actually maintain the models. Classical methods first, evaluation discipline throughout, and a final-day capstone on a dataset your team brings.
- /1/ Framing problems: what is and isn't an ML problem
- /2/ Feature engineering, leakage, and the bias-variance tradeoff
- /3/ Evaluation: cross-validation that respects time and groups
- /4/ When to escalate from logistic regression to gradient boosting to deep learning — and when not to
- /5/ Production handoff: model cards, monitoring, and the discipline of saying no