Databricks Machine Learning Associate Question 51
Single answerYou are using Databricks AutoML to train models for a regression problem. After the training process, you want to locate the source code for the best-performing model to understand how it was implemented and possibly customize it further. Where can you find the source code for the best model?
- A
In the 'Best Model' folder within the experiment's MLflow artifacts
- B
In the 'Notebooks' workspace under a generated notebook named after the experiment
- C
In the 'Feature Store' associated with the experiment
- D
In the 'Generated Code' section of the AutoML experiment's output
Show answer and explanation
Correct answer: D
Explanation
Databricks AutoML generates Python source code for each model it trains, including the best-performing model. This source code is stored in the 'Generated Code' section of the AutoML experiment's output. This allows users to review, understand, and modify the model's implementation if needed. Other locations, such as the MLflow artifacts or Feature Store, do not provide the actual source code for the best model.
- A. Incorrect.
The 'Best Model' folder within MLflow artifacts contains serialized model files and related artifacts but not the source code.
- B. Incorrect.
The 'Notebooks' workspace may contain other notebooks but does not store the AutoML-generated source code for the best model.
- C. Incorrect.
The 'Feature Store' is used to manage and track features for machine learning models, not to store the source code for the best model.
- D. Correct.
The 'Generated Code' section of the AutoML experiment's output contains the Python source code used to train the best model, making it the correct location.