Google Professional Machine Learning Engineer Question 90
Select 3Google Cloud PlatformYour organization stores large amounts of data in Google Cloud Storage and BigQuery. You have been tasked with building a machine learning pipeline to predict customer churn. The data consists of unstructured logs in Cloud Storage and structured customer profiles in BigQuery. Which of the following steps should you take to preprocess and explore the data before training the model?
- A
Use Dataflow to transform the unstructured logs in Cloud Storage into a structured format suitable for analysis.
- B
Directly train the model on the raw unstructured logs from Cloud Storage without any preprocessing.
- C
Use SQL queries in BigQuery to clean and normalize the structured customer profiles.
- D
Leverage Dataprep to visually explore and clean both the unstructured and structured datasets.
- E
Export the data from BigQuery and Cloud Storage to your local machine for preprocessing using custom Python scripts.
Show answer and explanation
Correct answers: A, C, D
Explanation
To preprocess and explore organization-wide data stored in Google Cloud Storage and BigQuery, it is important to use scalable and cloud-native tools such as Dataflow, BigQuery SQL, and Dataprep. These tools allow you to handle large datasets effectively, transform unstructured data into structured formats, and clean and normalize data for machine learning. Avoid relying on manual or local preprocessing methods, as they do not leverage the scalability and efficiency of cloud infrastructure.
- A. Correct.
Dataflow is a suitable tool for transforming large unstructured datasets in Cloud Storage into a structured format. This step is essential for making the data usable for machine learning.
- B. Incorrect.
Training the model on raw unstructured logs without preprocessing is not recommended, as unprocessed data is unlikely to provide meaningful features for the model.
- C. Correct.
Using SQL queries in BigQuery to clean and normalize structured data is an efficient and scalable approach to prepare the data for machine learning.
- D. Correct.
Dataprep is a powerful tool for visually exploring and cleaning datasets stored in BigQuery and Cloud Storage. This step helps ensure data quality before model training.
- E. Incorrect.
Exporting data to your local machine for preprocessing is inefficient and not scalable, especially when working with large datasets in the cloud.