Question

You are training a scikit-learn classifier using a command job in Azure Machine Learning. In your training script you already call mlflow.autolog(), but the business requires a custom metric — the F2 score computed on a held-out validation set — to be tracked in each MLflow run so you can later compare it across jobs in the studio UI. What should you add to the training script to capture this custom metric?

Accepted Answer

Call mlflow.log_metric("f2_score", f2_value) after computing the F2 score on the validation set.

Answer

Call mlflow.log_param("f2_score", f2_value) so the value is stored with the run.

Answer

Rely on mlflow.autolog() to automatically detect and log the F2 score with no additional code.

Answer

Add the F2 score to the model signature using mlflow.models.infer_signature() before registering the model.