Use the Snowflake Model Registry SNOWPRO-ADVANCED-DATA-SCIENTIST Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A startup is new to Snowflake and wants to manage their machine learning models centrally. They have a mix of models trained in different frameworks (scikit-learn, XGBoost) and want to deploy them for inference within Snowflake. What is the primary benefit of using the Snowflake Model Registry for this purpose?

   • AIt provides a centralized repository for models with versioning, metadata, and deployment capabilities.
   • BIt eliminates the need for separate training environments.
   • CIt provides a graphical interface for model training.
   • DIt automatically converts models to SQL for faster inference.
   Reveal Solution Discussion Select an answer first
2. 17foundation · easy

   How does the Snowflake Model Registry integrate with Snowpark ML?

   • AThe registry replaces Snowpark ML for model training
   • BSnowpark ML models can be directly registered into the registry
   • CSnowpark ML models cannot be registered in the registry
   • DThe registry only supports models trained outside Snowflake
   Reveal Solution Discussion Select an answer first
4. 18foundation · easy

   Which SQL function is used to deploy a registered model for inference in Snowflake?

   • APREDICT
   • BCREATE MODEL
   • CMODEL_INFERENCE
   • DDEPLOY_MODEL
   Reveal Solution Discussion Select an answer first
5. 19application · medium

   A data scientist has registered a model in the Snowflake Model Registry and wants to deploy it for real-time inference within Snowflake using SQL. They have a table 'features' with the required input columns. What is the correct way to invoke the model to get predictions?

   • AExport the features table to a CSV and use a Python script to run inference, then load the results back.
   • BUse the CALL statement with the model name and the features table as an argument.
   • CUse the model's prediction function in a SELECT statement, passing the feature columns as arguments.
   • DUse the MODEL_DEPLOY procedure to deploy the model as a view, then query the view.
   Reveal Solution Discussion Select an answer first
6. 20application · medium

   A data analyst needs to find all models in the Snowflake Model Registry that were created by the 'fraud_team' and have a tag 'status=production'. They want to use SQL to query the registry. Which approach should they use?

   • AUse the Python API to list models and filter in Python, but this is not SQL.
   • BQuery the ACCOUNT_USAGE.MODEL_REGISTRY view, which includes all metadata.
   • CQuery the MODEL_REGISTRY information schema views, filtering on the owner and tag columns.
   • DUse the SHOW MODELS command and filter the results in the client.
   Reveal Solution Discussion Select an answer first