AHIC Data Analytics and Data Management 2 — Questions and Answers
Question 1: Which data warehouse architecture pattern stores pre-aggregated data in subject-oriented structures to optimize query performance?
- Data vault
- Star schema (Correct answer)
- Snowflake schema
- Third normal form
Correct answer: Star schema
Star schema stores fact tables surrounded by denormalized dimension tables, enabling fast OLAP query performance through pre-aggregation.
Question 2: A health system wants to predict patient readmission risk. Which machine learning approach is most appropriate when labeled outcome data is available?
- Unsupervised clustering
- Reinforcement learning
- Supervised classification (Correct answer)
- Association rule mining
Correct answer: Supervised classification
Supervised classification uses labeled historical data (readmitted vs. not) to train a model that predicts outcomes for new patients.
Question 3: In healthcare data governance, which role is primarily responsible for defining data quality standards and authorizing data access policies?
- Data custodian
- Data steward (Correct answer)
- Database administrator
- Chief Information Officer
Correct answer: Data steward
Data stewards own the business definition, quality standards, and access policies for their assigned data domains.
Question 4: An analyst receives a dataset where patient age is recorded as '999' for unknown values. This is an example of which data quality dimension issue?
- Lack of timeliness
- Lack of completeness
- Lack of validity (Correct answer)
- Lack of uniqueness
Correct answer: Lack of validity
Validity refers to data conforming to defined formats, ranges, and rules; '999' for age violates the valid range for a real patient age.
Question 5: Which SQL window function calculates a running total without collapsing rows into a single aggregate result?
- GROUP BY with SUM
- SUM() OVER (ORDER BY ...) (Correct answer)
- HAVING SUM(...)
- DISTINCT COUNT
Correct answer: SUM() OVER (ORDER BY ...)
SUM() OVER (ORDER BY ...) is a window function that computes a cumulative sum while retaining each individual row in the result set.
Question 6: The FAIR data principles require that health data be Findable, Accessible, Interoperable, and Reusable. Which principle is directly addressed by assigning a persistent, globally unique identifier to each dataset?
- Accessible
- Interoperable
- Reusable
- Findable (Correct answer)
Correct answer: Findable
Assigning persistent unique identifiers (e.g., DOIs or URIs) makes datasets Findable by both humans and machines.
Question 7: Which ETL technique is most efficient when only changed records since the last extraction need to be loaded into a data warehouse?
- Full load
- Incremental load using change data capture (Correct answer)
- Truncate and reload
- Parallel bulk insert
Correct answer: Incremental load using change data capture
Change data capture (CDC) identifies and extracts only records modified since the last load, minimizing processing time and system load.
AHIC Data Analytics and Data Management 2
Which data warehouse architecture pattern stores pre-aggregated data in subject-oriented structures to optimize query performance?