Google Professional Machine Learning Engineer Question 78
Select 3Google Cloud PlatformYou are working for a retail company that wants to predict customer churn using their historical customer transactions dataset stored in BigQuery. The dataset includes numerical and categorical features such as total purchases, last purchase date, customer demographic information, and subscription type. You decide to use AutoML Tables to build the model. Which of the following steps should you perform to prepare and train your model?
- A
Export the dataset from BigQuery to Cloud Storage as CSV files before using AutoML Tables.
- B
Ensure your dataset has a target column for churn prediction and split it into training, validation, and test sets.
- C
Use the BigQuery connector in AutoML Tables to directly import data from BigQuery.
- D
Manually encode categorical features into numerical values using one-hot encoding before training.
- E
Specify the target column and configure relevant training options such as evaluation metrics in AutoML Tables.
Show answer and explanation
Correct answers: B, C, E
Explanation
To use AutoML Tables for tabular data, you need to ensure the dataset is prepared correctly with a target column and appropriate splits (training, validation, and test sets). AutoML Tables can directly integrate with BigQuery, so exporting the data is unnecessary. Additionally, AutoML Tables automatically handles feature engineering, including encoding categorical features, so there is no need for manual preprocessing. Finally, you must specify the target column and configure training options to initiate the model training process.
- A. Incorrect.
Incorrect: AutoML Tables supports direct integration with BigQuery, so exporting the dataset to Cloud Storage is unnecessary.
- B. Correct.
Correct: AutoML Tables requires a target column for supervised learning tasks, and splitting the data into training, validation, and test sets is a key step in preparing the data.
- C. Correct.
Correct: AutoML Tables provides a direct BigQuery connector, allowing users to import data without exporting it to another storage format.
- D. Incorrect.
Incorrect: AutoML Tables automatically handles categorical feature encoding, so manual one-hot encoding is not required.
- E. Correct.
Correct: Specifying the target column and configuring training options, such as evaluation metrics, is necessary to train the model using AutoML Tables.