Question

You train a scikit-learn classifier inside an Azure Machine Learning job and use MLflow to track the run. You want the trained model to be logged with the run so that it can later be registered as a model version and deployed without writing a custom scoring script. Which MLflow API call should you use in your training script to accomplish this?

Accepted Answer

mlflow.sklearn.log_model() to log the model with its flavor and inferred signature

Answer

mlflow.log_artifact() to upload the serialized pickle file of the model

Answer

mlflow.log_param() to record the model hyperparameters

Answer

mlflow.log_metric() to record the model's accuracy score