Databricks Certification Practice Test 12
Practice Test 12
Databricks Certification Practice Test 12
Practice Test 12
Insightful Saga — Modern Data Engineering Certification Preparation
Question 261
## Question 261 A retail company has developed a fraud detection model. Training accuracy remains above 95%, but production accuracy falls every month even though no code changes were deployed. What is the MOST likely explanation?
Question 262
## Question 262 A healthcare company retrains models every week. Management wants assurance that newly trained models outperform currently deployed models before promotion. What should be implemented?
Question 263
## Question 263 A recommendation system requires feature values that are less than 30 seconds old. What architecture is most suitable?
Question 264
## Question 264 An ML team discovers that training data contains significantly fewer examples of rare fraud events than normal transactions. What ML challenge exists?
Question 265
## Question 265 A company wants to understand which input features most influenced a model prediction. What capability should be implemented?
Question 266
## Question 266 A retailer trains a model using historical sales data. Engineers later discover future sales information accidentally existed in training data. What issue occurred?
Question 267
## Question 267 A data scientist wants to automatically search thousands of parameter combinations. What capability is most appropriate?
Question 268
## Question 268 An enterprise AI platform wants standardized model packaging regardless of programming language. What should be prioritized?
Question 269
## Question 269 A manufacturing company wants to predict machine failures before they occur. What ML category best fits this use case?
Question 270
## Question 270 A bank requires every prediction generated by a model to be reproducible years later. What is most critical?
Question 271
## Question 271 A company wants to detect previously unseen fraudulent behavior patterns. Which approach is most appropriate?
Question 272
## Question 272 A model performs well in North America but poorly in Asia. What should engineers investigate first?
Question 273
## Question 273 A company wants an AI model deployment strategy where only 10% of traffic initially reaches the new model. What should be implemented?
Question 274
## Question 274 An ML platform team wants to compare production predictions with actual outcomes over time. What should be measured?
Question 275
## Question 275 A company has hundreds of trained models. Management wants visibility into which models are actively used. What should be tracked?
Question 276
## Question 276 A recommendation model requires user clickstream behavior within seconds. Which data pattern is required?
Question 277
## Question 277 A support organization wants AI systems to explain confidence for each answer generated. What should be exposed?
Question 278
## Question 278 A team trains multiple models using the same customer features. Management wants one approved version of features. What principle should be applied?
Question 279
## Question 279 A logistics company predicts shipment delays. Business users want predictions before delays occur, not after. What type of analytics is being requested?
Question 280
## Question 280 A model generates excellent results but requires 30 seconds per prediction. The business requires sub-second responses. What should be optimized?
Question 281
## Question 281 A company notices predictions change after retraining despite identical parameters. What should be investigated?
Question 282
## Question 282 An AI assistant must answer customer questions in 25 different languages. What capability becomes critical?
Question 283
## Question 283 A company wants AI applications to learn from customer feedback over time. What process should exist?
Question 284
## Question 284 A fraud detection model consumes 10,000 features. Only 50 significantly affect outcomes. What engineering task may improve performance?
Question 285
## Question 285 A company wants every feature used in production predictions to be traceable back to source systems. What should be established?
Question 286
## Question 286 A model serving endpoint experiences sudden traffic spikes. What architecture principle helps absorb demand?
Question 287
## Question 287 An AI application produces responses that vary significantly for identical prompts. What should engineers investigate?
Question 288
## Question 288 A company wants to identify whether model performance differs across age groups. What discipline is being applied?
Question 289
## Question 289 A financial institution wants to evaluate risk before deploying a new model. What practice should be performed?
Question 290
## Question 290 A chatbot must access company policies, legal agreements, and procedure manuals from multiple repositories. What architecture is most suitable?
Question 291
## Question 291 A model predicts customer churn. Business leaders want an explanation for the top reasons contributing to churn. What should be implemented?
Question 292
## Question 292 A company wants to compare model quality across multiple business units using a common evaluation standard. What should be established?
Question 293
## Question 293 A model retraining workflow consumes significant compute resources. Management wants to train only when statistically necessary. What strategy should be adopted?
Question 294
## Question 294 An enterprise wants to identify duplicate AI models solving the same business problem. What capability would help?
Question 295
## Question 295 A recommendation system serves millions of customers globally. The business requires uninterrupted availability during deployments. What approach should be used?
Question 296
## Question 296 A company wants to ensure AI-generated content follows corporate brand guidelines. What should be implemented?
Question 297
## Question 297 A data scientist wants to quickly identify which training runs produced the highest business value. What should be tracked alongside technical metrics?
Question 298
## Question 298 A GenAI platform serves thousands of users. Management wants visibility into token consumption and operational costs. What should be monitored?
Question 299
## Question 299 An organization wants a single framework governing ML models, GenAI applications, feature definitions, evaluations, and deployments. What should be established?
Question 300
## Question 300 A global enterprise is building a Lakehouse AI platform expected to support analytics, machine learning, GenAI, AI agents, and future AI workloads. What architectural characteristic is MOST important?
An extensible architecture allows AI platforms to evolve as new models, frameworks, and business requirements emerge. ---