An ML engineer is inspecting a classification dataset before training a model in Amazon SageMaker and suspects a large class-label imbalance that could bias the model. Which pre-training metric should the engineer use to confirm class imbalance?

The Difference in Proportions of Labels (DPL) is the correct metric because it directly quantifies class imbalance by measuring the difference in the proportion of samples belonging to different classes. A large DPL value indicates a significant imbalance, which can lead to biased models that perform poorly on minority classes. Mean Squared Error (MSE) is a regression metric, not suitable for classification or imbalance detection. Silhouette score is used for evaluating clustering algorithms. Structural Similarity Index Measure (SSIM) is used for measuring the similarity between two images.