classical-mlfoundations
Evaluation metrics
Accuracy, AUC, F1, BLEU, NDCG — choosing and computing the right metric for each ML task.
Уровни глубины
L0Intro~1ч
Knows accuracy and why it can be misleading on imbalanced data.
L1Basics~8ч
Computes precision, recall, F1, AUC-ROC, confusion matrix; selects metric by task.
L2Working~15ч
Uses ranking metrics (NDCG, MAP, MRR); applies calibration; selects thresholds for business objectives.
L3Advanced~25ч
Designs custom loss functions aligned with business metrics; analyses statistical significance of metric differences.
L4Research~50ч
Develops new evaluation protocols or adversarial robustness benchmarks.
Ресурсы
L0 — Intro
L1 — Basics
L2 — Working