hardMultiple Choice
Custom Sentiment Analysis with AutoML Natural Language
A financial institution wants to use Natural Language API for sentiment analysis on customer feedback, but the domain-specific language (e.g., 'bullish', 'bearish') is not correctly classified. They have 200 labeled examples. Which approach minimizes coding effort while improving accuracy?
Quick Answer
The answer is to use AutoML Natural Language to train a custom model. This is correct because AutoML Natural Language leverages transfer learning from Google’s pre-trained models, allowing you to build a custom sentiment model with few labeled examples—here, just 200—without writing any code. The domain-specific terms like 'bullish' and 'bearish' are misclassified by the general Natural Language API because its pre-trained model lacks exposure to financial jargon; AutoML fine-tunes on your labeled data to adapt to this unique vocabulary and sentiment patterns, directly improving accuracy. On the Google Professional Machine Learning Engineer exam, this scenario tests your understanding of when to use AutoML versus the pre-trained API—a common trap is assuming you need to write custom code or use a different service like Vertex AI Workbench, but AutoML minimizes coding effort by handling model training and evaluation automatically. Memory tip: think “AutoML for auto-magic fine-tuning” when you have under 1,000 labeled examples and need domain-specific sentiment.
⚠ Common exam trap
Google Cloud often tests the misconception that the Natural Language API supports custom dictionaries or rule-based overrides, when in fact it only offers a fixed pre-trained model, making AutoML the correct low-code path for domain adaptation.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Use AutoML Natural Language to train a custom model
AutoML Natural Language enables you to train a custom model on your 200 labeled examples without writing code, directly improving accuracy for domain-specific terms like 'bullish' and 'bearish'. This approach leverages transfer learning from Google's pre-trained models, minimizing coding effort while adapting to your unique vocabulary and sentiment patterns.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Submit a feature request to Google for domain-specific terms
Why it's wrong here
A feature request to Google delivers no accuracy improvement within the project timeline, since roadmap decisions are outside the customer's control. It is tempting when a managed API lacks domain coverage, but the scenario supplies 200 labelled examples precisely so AutoML Natural Language can train a domain-specific model now.
- ✗
Create a custom sentiment dictionary and pass it to the Natural Language API
Why it's wrong here
The Natural Language API accepts no custom sentiment dictionary parameter; its model weights are fixed, so supplying domain terms changes nothing. It is tempting because lexicon overrides work in tools such as VADER or spaCy, but here the API exposes no such hook, leaving AutoML as the supported route.
- ✗
Build a custom TensorFlow model for sentiment
Why it's wrong here
Building a TensorFlow model demands substantial coding, training infrastructure and tuning, contradicting the minimal-effort requirement despite only 200 labelled examples. It is tempting when labelled data exists and accuracy matters, but AutoML Natural Language classification trains on those examples with far less code.
- ✓
Use AutoML Natural Language to train a custom model
Why this is correct
AutoML Natural Language trains a custom sentiment model on the 200 labelled examples through the console, learning domain terms such as 'bullish' and 'bearish' that the pretrained Natural Language API misclassifies. This meets the minimal-coding constraint while adapting to the financial vocabulary.
Go deeper
Related to this question
About these practice questions
Courseiva writes every PMLE question from scratch — 775 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. Learn why practice questions differ from exam dumps →
Same concept, more angles
1 more way this is tested on PMLE
These questions test the same concept from different angles. Work through them to make sure you can recognise it however the exam phrases it.
Variation 1. A small business wants to build a sentiment analysis model for customer reviews without writing any code. They have a small labeled dataset with 500 positive and 500 negative reviews. Which Google Cloud service should they use?
easy- ✓ A.AutoML Natural Language
- B.Natural Language API
- C.Vertex AI custom training with PyTorch
- D.BigQuery ML with logistic regression
Why A: AutoML Natural Language is the correct choice because it allows the business to train a custom sentiment analysis model using their own labeled dataset without writing any code. It provides a low-code interface for uploading data, training, and deploying the model, which aligns with the requirement of no coding and a small labeled dataset.
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This PMLE practice question is part of Courseiva's free Google Cloud certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the PMLE exam.