Imbalanced Classification — Google Cloud ML Engineer Practice Questions

Imbalanced classification refers to scenarios where one class in the training data is vastly more frequent than another, such as fraud detection where fraudulent transactions represent a tiny fraction of all transactions. Standard accuracy becomes a misleading metric in this setting because a model that always predicts the majority class can achieve high accuracy while being completely useless. The Google Cloud ML Engineer exam tests strategies for handling class imbalance, including oversampling minority classes, undersampling majority classes, adjusting class weights in Vertex AI and BigQuery ML training configurations, and choosing evaluation metrics like precision, recall, F1-score, and AUC-ROC that are more informative than raw accuracy.

Free questions on imbalanced classification

More imbalanced classification questions in the full bank

• In a highly imbalanced classification task, why is stratified k-fold better than regular k-fold? Unlock answer & explanation →