Free Google Cloud-Digital-Leader Actual Exam Questions - Question 2 Discussion
busy season. They do not have a data science team, and want to use their in-house database
administration skills to create a machine learning model. What should the organization do?
C Using AutoML could be a good fit since it handles the model building without deep data science skills, and it works well with labeled data that they likely have from past rentals.
B/D? If they have SQL skills, BigQuery ML (B) is straightforward but only if data’s in BigQuery. Otherwise, using pre-trained APIs (D) might save hassle since no model building is needed.
Option C also makes sense since AutoML handles the heavy lifting and requires less technical setup, perfect for a team without dedicated data scientists. It’s user-friendly for non-experts and uses labeled data effectively.
Makes sense to skip custom TensorFlow—their skills align well with SQL in BigQuery ML, so B.
It’s B for me too. Since they have database admin skills, using BigQuery ML to build models with SQL seems like the smoothest path without needing heavy data science expertise. Unlike option C, AutoML might be more of a black box and could require more setup or data prep than they’re ready for. Also, pre-trained APIs (D) usually cover generic tasks, not something specific like property popularity prediction here. The key question is if they can get their data into BigQuery easily, but assuming that’s manageable, B fits their skill set best.
They probably won’t want to handle the complexity of custom TensorFlow training (A). Since they have database skills, BigQuery ML (B) lets them build models using SQL, which fits their in-house expertise better than AutoML or APIs.
D imo, integrating pre-trained APIs is the quickest way without needing much ML expertise. It’s less about building models and more about leveraging existing solutions, which fits a non-DS team better.
B seems best since they have database skills and BigQuery ML lets you build models with SQL.