subtopic

Classification metrics from the confusion matrix

Write evaluate_classifier(y_true, proba, threshold) returning confusion matrix, accuracy, precision, recall, F1, F2, cost. Concepts: TP/FP/TN/FN; accuracy; precision; recall; F1; F-beta; classification report; business cost Project A: Main metrics: Bad recall, Bad precision, F2, cost = 5·FN + FP. Project B: Not applicable (classification-only method). (Optional: a 'peak-day alert' classifier would use the same harness.)

Builds on: B18 · Build: Spam Classifier with Probabilities (sigmoid, softmax, cross-entropy)