Practice C14-013–024
C14-013 — Machine Learning & Mining · Foundational · Recall
A spam classifier is trained from emails already labeled spam or not spam. Which learning type best fits? - A. Supervised learning. - B. Unsupervised learning. - C. Reinforcement learning. - D. Association mining.
C14-014 — Machine Learning & Mining · Standard · Apply
Marketing has no predefined customer segments and wants to discover natural groups from behavior. Which approach is strongest? - A. Supervised learning. - B. Unsupervised learning using clustering. - C. Reinforcement learning. - D. Prescriptive analytics.
C14-015 — Machine Learning & Mining · Standard · Distinguish
Which statement best distinguishes association from clustering? - A. Association requires labels; clustering never uses data values. - B. Association predicts the future; clustering describes the past. - C. Association finds elements/events that occur or relate together; clustering groups similar cases or observations. - D. Association is Data Quality; clustering is Metadata.
C14-016 — Machine Learning & Mining · Standard · Understand
Why is simple keyword counting often inadequate for sentiment analysis? - A. Sentiment can only be measured from numeric transaction data. - B. Keyword counts are always too slow for Big Data. - C. Sentiment analysis is a form of Master Data Management. - D. Meaning depends on linguistic context, negation, and how words are used together, so isolated counts can misclassify sentiment.
C14-017 — Machine Learning & Mining · Difficult · Distinguish
A retailer wants to discover products that commonly appear together in baskets. Which technique is most direct? - A. Association analysis / data mining. - B. Clustering. - C. Prescriptive analytics. - D. Data reduction.
C14-018 — Machine Learning & Mining · Expert · Apply
An algorithm improves a sequence of actions based on feedback indicating whether each action moves it closer to a defined goal. Which learning approach best fits? - A. Supervised learning. - B. Reinforcement learning. - C. Unsupervised learning. - D. Descriptive analytics.
C14-019 — Strategy & Sources · Foundational · Recall
Which should come first in Chapter 14: selecting the Big Data platform or defining strategy/business needs? - A. Select the platform first so it can define available use cases. - B. Acquire every available source first. - C. Define strategy and business needs first. - D. Train a model first and define the business benefit only if accuracy is high.
C14-020 — Strategy & Sources · Standard · Understand
Which source-selection characteristic asks whether a data set is at the right level of detail for the intended analysis? - A. Velocity. - B. Veracity. - C. Volatility. - D. Granularity.
C14-021 — Strategy & Sources · Standard · Apply
A social feed is very large and inexpensive, but the provider frequently changes field definitions and cannot explain how the sample is constructed. What is the best next step? - A. Evaluate and profile the source for meaning, reliability, consistency, population bias, and Metadata before selection/ingestion. - B. Ingest it first because profiling is more accurate after full integration. - C. Use it only in the speed layer so reliability matters less. - D. Treat the provider's field names as sufficient Metadata.
C14-022 — Strategy & Sources · Difficult · Apply
A customer model is trained only on mobile-app users but will be applied to all customers. Which risk should be addressed before relying on the model? - A. Volume risk because the app data may be too small. - B. Population/source-selection bias caused by systematically excluding non-app customers. - C. A data-swamp risk because the source is mobile. - D. A serving-layer risk because the prediction is enterprise-wide.
C14-023 — Strategy & Sources · Difficult · Apply
Two individually anonymous sources are combined and the resulting records can identify a small group of people. What is the strongest Chapter 14 concern? - A. Only Data Quality, because the join may be inaccurate. - B. Only Metadata, because source provenance changed. - C. Privacy/security risk from recombination. - D. Only Volume, because combining sources increases row count.
C14-024 — Strategy & Sources · Expert · Understand
A source is statistically reliable and complete, but nobody can document its origin, meaning changes, lineage, or intended use. Which gap is primary before enterprise reuse? - A. Data Quality only. - B. Prescriptive analytics. - C. MPP architecture. - D. Metadata Management/context gap.