Google Professional Machine Learning Engineer Question 3
Single answerGoogle Cloud PlatformYou are tasked with building a machine learning model for predicting customer churn using BigQuery ML. The dataset is stored in BigQuery and contains customer behavior metrics such as 'last_purchase_date', 'average_monthly_spending', and 'total_logins'. You need a model that provides interpretable predictions and supports SQL-based feature engineering directly within BigQuery. Which model type should you choose for this task?
- A
Logistic Regression
- B
Deep Neural Network
- C
k-means Clustering
- D
AutoML Tables
Show answer and explanation
Correct answer: A
Explanation
BigQuery ML supports building interpretable models like Logistic Regression directly within BigQuery using SQL. For a binary classification task such as churn prediction, Logistic Regression is a natural fit. It allows you to leverage SQL for feature engineering and ensures the model is interpretable, which is critical for understanding the reasons behind customer churn.
- A. Correct.
Logistic Regression is suitable for binary classification tasks like churn prediction. It is interpretable and natively supported by BigQuery ML, making it a good choice for this scenario.
- B. Incorrect.
Deep Neural Network is a powerful model for complex tasks, but it is less interpretable and not the best choice for a straightforward binary classification task with structured data.
- C. Incorrect.
k-means Clustering is an unsupervised learning algorithm used for grouping data points into clusters. It is not suitable for a supervised task like churn prediction.
- D. Incorrect.
AutoML Tables is a managed solution for building machine learning models with minimal effort, but it operates outside BigQuery ML and does not allow direct SQL-based feature engineering within BigQuery.