問題文
A data scientist needs the predicted class for rows in a table and the business problem is a standard binary classification on tabular columns. The team has no Python environment set up and wants the shortest path. Which option should be considered first?
選択肢
- A custom model trained with a Python library, set up on a laptop and packaged as a model artifact, because only a custom model can produce class predictions.
- A built-in classification capability invoked from sql, because it trains and predicts without requiring a Python environment.
- An external function calling a hosted service, with a network rule and Snowflake access integration, because that avoids setting up anything inside the account.
- A container-based service, with a compute pool, because inference always requires a container of its own whenever the team has no Python environment set up anywhere.