The correct answer is C. experiment_dir. This parameter allows you to specify a directory where AutoML will store the generated notebooks and experiment files, enabling a structured workspace for easy access and analysis.

How to use experiment_dir:

from databricks import automl
# Assuming your regression dataset is a Spark DataFrame named ‘data‘
automl_model = automl.regression(
    data=data,
    target_col=“target_column“,  # Column containing the target values to predict
    experiment_dir=“/dbfs/my_experiments/regression_run“,  # Specify the experiment directory
    # … other AutoML parameters
)

from databricks import automl
# Assuming your regression dataset is a Spark DataFrame named ‘data‘
automl_model = automl.regression(
    data=data,
    target_col=“target_column“,  # Column containing the target values to predict
    experiment_dir=“/dbfs/my_experiments/regression_run“,  # Specify the experiment directory
    # … other AutoML parameters
)

Incorrect Options:

• feature_store_lookups: Manages feature lookups from a feature store, unrelated to experiment storage.
• exclude_cols: Excludes specific columns from consideration, not experiment storage.
• primary_metric: Specifies the primary metric for model evaluation, not experiment storage.

Key Points:

• Use experiment_dir for organized experiment management.
• Ensure the specified directory exists in your Databricks file system (DBFS).

Benefits of using experiment_dir:

• Enhanced organization and traceability.
• Easier collaboration and sharing of results.
• Ability to revisit experiments for analysis or model re-deployment.
• Facilitates tracking progress and comparing different AutoML runs.