A company wants to use machine learning models to predict customer churn. What should they prioritize in their dataset?
AIncreasing dataset size
BReducing feature dimensionality
CAdding irrelevant features
DBalancing class distribution
Explanation
Balancing class distribution is crucial for accurate predictions, while irrelevant features and dimensionality reduction might not help as much in this context.
Q192
You are configuring Google Cloud IAM roles. What happens if a user has multiple roles with different permissions?
AOnly the highest role applies
BAll roles are combined
COnly the lowest role applies
DPermissions are ignored
Explanation
All roles are combined, granting cumulative permissions, making option A incorrect as the highest does not override others.
Q193
Which service enables real-time data processing in Google Cloud?
ACloud Pub/Sub
BBigQuery
CDataflow
DCloud Storage
Explanation
Dataflow is specifically designed for real-time processing, while the others provide different functionalities.
Q194
A company needs to comply with GDPR for its machine learning datasets. What should it do?
AEncrypt the datasets at rest
BStore data in the U.S.
CRemove all personal identifiers
DLimiting data access to engineers only
Explanation
Removing personal identifiers is key for GDPR compliance, unlike options B and D which do not ensure full compliance.
Q195
You are configuring a TensorFlow model for deployment on Vertex AI. What should you ensure before deployment?
AModel is TensorFlow 2.x compatible
BModel must use Keras exclusively
CModel is optimized for CPU only
DModel training must be completed locally
Explanation
TensorFlow 2.x compatibility is required for seamless deployment, while other options impose unnecessary restrictions.
Q196
Which Google Cloud service is best for real-time data streaming?
APub/Sub
BBigQuery
CCloud Storage
DDataflow
Explanation
Pub/Sub is designed for real-time messaging, while the others are for batch processing or storage.
Q197
A company needs to organize ML models with version control. What should they use?
ACloud Source Repositories
BAI Platform Models
CDocker
DCloud Functions
Explanation
AI Platform Models includes built-in version control for ML models, unlike the other options.
Q198
You are configuring AutoML for customer sentiment analysis. What happens when training encounters insufficient labeled data?
ATraining stops immediately
BModel performance declines
COverfitting occurs
DData augmentation happens
Explanation
Insufficient labeled data typically leads to poor model performance, whereas overfitting requires careful tuning.
Q199
Which Google Cloud service is best for real-time data processing?
APub/Sub
BCloud Functions
CCloud Storage
DCompute Engine
Explanation
Pub/Sub enables real-time message processing; others are not optimized for that use case.
Q200
A company needs to store sensitive JSON data securely. What should they use?
ACloud Firestore
BCloud SQL
CCloud Storage with encryption
DPersistent Disk
Explanation
Cloud Storage with encryption provides secure data storage; others lack built-in encryption for sensitive data.