Google Professional Machine Learning Engineer Question 481
Select 3Google Cloud PlatformYou are deploying a machine learning model for a credit risk scoring application using Vertex AI. The business stakeholders require that the model’s predictions be explainable to ensure regulatory compliance. Which feature of Vertex AI Prediction can you use to provide explanations for model predictions, and what steps are necessary to enable it?
- A
Enable model explainability by configuring explanation metadata during model deployment.
- B
Integrate the Explainable AI feature of Vertex AI by specifying feature importance as part of the model training pipeline.
- C
Use SHAP or Integrated Gradients as explanation methods supported by Vertex AI Prediction.
- D
Manually calculate feature importance scores after deploying the model by analyzing the predictions.
- E
Ensure explanation configuration is included in the model’s endpoint settings when deploying it to Vertex AI.
Show answer and explanation
Correct answers: A, C, E
Explanation
Vertex AI provides model explainability features that help interpret predictions using explanation metadata and supported methods like SHAP or Integrated Gradients. To enable these features, explanation metadata must be configured during deployment, and explanation settings must be included in the endpoint configuration. This allows stakeholders to understand the model’s behavior and maintain compliance with regulatory requirements, especially in sensitive applications like credit risk scoring.
- A. Correct.
Correct: Explanation metadata must be configured during model deployment to enable model explainability in Vertex AI Prediction.
- B. Incorrect.
Incorrect: Feature importance is not specified during the training pipeline but is computed as part of the explanation process in Vertex AI.
- C. Correct.
Correct: Vertex AI Prediction supports explanation methods such as SHAP and Integrated Gradients to provide insights into model predictions.
- D. Incorrect.
Incorrect: Feature importance scores are automatically generated by Vertex AI when explainability is enabled, so manual computation is not required.
- E. Correct.
Correct: Explanation configuration must be added to the endpoint settings during deployment to enable explainability features in Vertex AI.