Databricks Machine Learning Associate Question 156
Select 3You are using Databricks to train a machine learning model and want to monitor key metrics such as mean squared error and accuracy during training. Which of the following steps would allow you to summarize and log these metrics using MLflow?
- A
Use the mlflow.log_metric() function to record metrics at each training iteration.
- B
Enable MLflow autologging to automatically capture and summarize metrics during the training process.
- C
Use the display() function in Databricks to visualize metrics directly, which automatically logs them to MLflow.
- D
Manually create a Pandas DataFrame to store metrics and log it as an artifact in MLflow.
- E
Log a summary of the metrics at the end of training using mlflow.log_metric() for each key metric.
Show answer and explanation
Correct answers: A, B, E
Explanation
To summarize and monitor key metrics during training in a Databricks environment, you can use mlflow.log_metric() to log individual metrics manually or enable MLflow autologging for automatic logging and summarization. At the end of training, you should also log a summary of critical metrics for comprehensive tracking. Visualization tools like display() and storing metrics in artifacts do not directly contribute to logging summaries in MLflow.
- A. Correct.
Correct. The mlflow.log_metric() function is used to record individual metrics, which can be summarized over iterations during training.
- B. Correct.
Correct. MLflow autologging simplifies the process by automatically capturing and summarizing metrics for supported frameworks.
- C. Incorrect.
Incorrect. While the display() function helps with visualization in Databricks notebooks, it does not automatically log or summarize metrics into MLflow.
- D. Incorrect.
Incorrect. Although you can store metrics in a Pandas DataFrame, this approach does not directly log metrics as MLflow summaries. Artifacts are different from logged metrics.
- E. Correct.
Correct. Logging a summary of metrics at the end of training ensures key metrics are recorded in MLflow for later analysis.