ML//Evaluation

How you measure whether a model actually works: metrics, splits, controls, and the ways a number can lie. Groups PR-AUC, ROC-AUC, cross-validation, data leakage, negative control, and class imbalance.

How you measure whether a model actually works: metrics, splits, controls, and the ways a number can lie. Groups PR-AUC, ROC-AUC, cross-validation, data leakage, negative control, and class imbalance.