Discuss Google Professional Data Engineer Exam Topic 1 Question 112

A data scientist has created a BigQuery ML model and asks you to create an ML pipeline to serve predictions. You have a REST API application with the requirement to serve predictions for an individual user ID with latency under 100 milliseconds. You use the following query to generate predictions: SELECT predicted_label, user_id FROM ML.PREDICT (MODEL 'dataset.model', table user_features). How should you create the ML pipeline?

AAdd a WHERE clause to the query, and grant the BigQuery Data Viewer role to the application service account.

BCreate an Authorized View with the provided query. Share the dataset that contains the view with the application service account.

CCreate a Cloud Dataflow pipeline using BigQueryIO to read results from the query. Grant the Dataflow Worker role to the application service account.

DCreate a Cloud Dataflow pipeline using BigQueryIO to read predictions for all users from the query. Write the results to Cloud Bigtable using BigtableIO. Grant the Bigtable Reader role to the application service account so that the application can read predictions for individual users from Cloud Bigtable.

Suggested Answer: D

Get Premium Professional Data Engineer Questions as Interactive Web-Based Practice Test or PDF Real Exam Format Detailed Explanations Instant Access Get Your Practice Test

23 comments · newest first

Milly 10 months ago

D sounds complicated, not sure if it's necessary for individual predictions.

Kate 10 months ago

C is a solid choice, but does it really meet the latency requirement?

Malcom 11 months ago

I think B is better for security with the Authorized View.

Fabiola 11 months ago

I agree with B, sharing the dataset is a smart move!

Tayna 11 months ago

Option A seems straightforward, just filter by user ID.

Clarinda 11 months ago

I feel like using Bigtable could be overkill for just one user’s prediction, but I remember it being mentioned as a way to handle larger datasets efficiently.

Verlene 11 months ago

I practiced a similar question about using Cloud Dataflow, but I can't recall if it's the best option here since we need predictions for a single user.

Katina 12 months ago

I think creating an Authorized View could be a good approach since it simplifies access control, but I’m not entirely sure how it impacts performance.

Jesusa 12 months ago

I remember we discussed the importance of latency in serving predictions, but I'm not sure if adding a WHERE clause is enough to meet the 100 ms requirement.

Jennifer 12 months ago

This is a tricky one. I'm leaning towards Option C with the Cloud Dataflow pipeline. That way, I can control the entire data processing flow and ensure the latency requirements are met. The Dataflow Worker role should give the application the necessary access.

Melinda 1 year ago

Okay, I think I've got a strategy here. Option B with the Authorized View seems like the best approach. That way, I can grant the necessary permissions to the application service account and the query will be optimized for low latency predictions.

Gail 1 year ago

Hmm, I'm a bit confused. The question mentions a REST API application, but it's not clear how that fits into the pipeline. I'll need to think through the different options and how they address the latency requirement.

Frederic 1 year ago

This seems like a straightforward question, but I want to make sure I understand the requirements correctly. The key is to create a pipeline that can serve predictions for individual users with low latency.

Precious 1 year ago

I think option D is the way to go. Using Cloud Dataflow to read predictions for all users and writing them to Cloud Bigtable will allow for efficient access to individual user predictions.

Lashaun 1 year ago

I'm leaning towards option C. Creating a Cloud Dataflow pipeline using BigQueryIO seems like a scalable solution for serving predictions with low latency.

Denny 1 year ago

I disagree, I believe option B is the best choice. Creating an Authorized View will provide the necessary access to the application service account.

Leigha 1 year ago

You know, I was thinking the same thing as Avery. The Dataflow pipeline seems like the best way to ensure low latency and high performance, even with a large dataset. Option C is my pick.

Avery 1 year ago

Hmm, I'm not sure about that. What if the dataset is really large? Won't that lead to performance issues? I'd go with Option C and use Dataflow to handle the reading and processing.

Amina 1 year ago

↳ replying to Avery

I agree, using Dataflow for reading and processing will help with performance issues.

Shawna 1 year ago

↳ replying to Avery

Option C seems like the best choice. Dataflow can handle large datasets efficiently.

Ceola 1 year ago

I think we should go with option A. Adding a WHERE clause to the query seems like the most efficient way to serve predictions for individual user IDs.

Marge 1 year ago

Option B is the way to go. Creating an Authorized View and sharing the dataset with the application service account is the most efficient and scalable solution here.

Dalene 1 year ago

↳ replying to Marge

B) Create an Authorized View with the provided query. Share the dataset that contains the view with the application service account.

Google Professional Data Engineer Exam - Topic 1 Question 112 Discussion