Google Professional Machine Learning Engineer Question 23
Select 2Google Cloud PlatformYou are working on a machine learning model to predict customer churn using BigQuery ML. Your dataset contains features such as 'customer_age', 'monthly_spend', 'region', and 'signup_date'. After training an initial model, you observe that some features are not contributing significantly to model performance. How can you use BigQuery ML to identify and select the most important features for your model?
- A
Use the ML.FEATURE_IMPORTANCE function to analyze the impact of each feature.
- B
Manually remove features with low correlation to the target variable based on exploratory data analysis.
- C
Use the ML.EXPLAIN_PREDICT function to understand feature importance for individual predictions.
- D
Enable the AUTO_CLASS_WEIGHTS option in BigQuery ML to automatically select the best features.
- E
Use BigQuery ML's feature selection by setting the model type to AUTO_ML.
Show answer and explanation
Correct answers: A, C
Explanation
BigQuery ML provides tools such as ML.FEATURE_IMPORTANCE to analyze the importance of features for a model. Additionally, ML.EXPLAIN_PREDICT can be used to gain insights into feature contributions for individual predictions. These tools are valuable for understanding which features are most impactful and should be retained for modeling, ensuring better model performance.
- A. Correct.
ML.FEATURE_IMPORTANCE is a built-in function in BigQuery ML that provides information about the contribution of each feature to the model, helping with feature selection.
- B. Incorrect.
While exploratory data analysis is useful, manually removing features based on correlation is not specific to BigQuery ML and may not always lead to the best model performance.
- C. Correct.
ML.EXPLAIN_PREDICT is used to understand how individual features contribute to specific predictions, which can indirectly help in identifying important features.
- D. Incorrect.
AUTO_CLASS_WEIGHTS is unrelated to feature selection; it deals with class imbalance in classification tasks.
- E. Incorrect.
BigQuery ML's AUTO_ML model selects hyperparameters but does not explicitly perform feature selection.