A data scientist must build a fraud detection model with far fewer fraud examples than legitimate ones, needs to check for bias before finalizing, and wants to move quickly with minimal operational overhead. Which option best meets these requirements?

The correct option addresses all requirements efficiently. SMOTE (Synthetic Minority Over-sampling Technique) can be applied within SageMaker Studio to handle the class imbalance directly, avoiding the overhead of EMR. SageMaker JumpStart provides pre-built models and solutions, enabling rapid development and deployment, which aligns with the need to move quickly. Amazon SageMaker Clarify is specifically designed for bias detection and explainability, making it the ideal tool for checking bias before finalizing the model. Incorrect options: Using Amazon EMR for SMOTE adds unnecessary operational overhead and complexity compared to performing it directly in SageMaker Studio. While A2I can be used for human review, SageMaker Clarify is better suited for algorithmic bias detection. This option also suggests EMR for SMOTE and A2I for bias review. EMR is overkill for SMOTE, and Clarify is the dedicated service for bias detection. SageMaker Model Monitor is primarily for monitoring model performance and data drift after deployment, not for checking bias before finalizing the model.