Build and Automate Pipelines CERTIFIED-DATA-SCIENCE-PRACTITIONER Practice Questions (Page 7)

Questions 31–35

1. 31application · medium

   A retail company's data engineering team currently uses hand-written Python scripts to extract sales data from a transactional database, transform it with custom logic, and load it into a cloud data warehouse nightly. The scripts are brittle, and the business wants to add new data sources (e.g., CRM, web analytics) quickly. The team wants to adopt a modern tool that centralizes transformation logic in version-controlled SQL while keeping the extraction and loading steps managed by a separate tool. Which tool combination best fits this requirement?

   • AUse dbt for transformations and Fivetran for extraction and loading.
   • BUse Fivetran for transformations and dbt for extraction and loading.
   • CUse Fivetran for everything, including transformation logic written in Python.
   • DUse dbt for both extraction and transformation, and keep the custom Python for loading.
   Reveal Solution Discussion Select an answer first
2. 32application · medium

   A media company is building a data lake to store raw clickstream data from its websites and mobile apps. The data arrives in JSON and Parquet formats. The data science team will later need to run complex transformations on this data for user behavior analysis. What is the most appropriate ETL strategy for this data lake?

   • AUse a traditional ETL approach: transform the JSON and Parquet into a relational schema before loading.
   • BLoad the data into a data warehouse instead of a data lake.
   • COnly load the Parquet files, as JSON is not suitable for a data lake.
   • DUse an ELT approach: load the raw JSON and Parquet files into the lake, and transform them later using tools like Spark or dbt.
   Reveal Solution Discussion Select an answer first
4. 33application · medium

   A retail company's data warehouse team is designing a new ETL process. The source system is a transactional database, and the target is a dimensional data warehouse. The team needs to ensure that data is cleansed, deduplicated, and conformed to standard dimensions (e.g., unified customer IDs) before it is loaded into the warehouse. In which stage of the ETL process should this work primarily occur?

   • AAfter loading, by running a separate data quality script.
   • BDuring the extraction stage, by pulling only the required fields.
   • CDuring the transformation stage, by applying business rules and data quality checks.
   • DDuring the loading stage, by using database constraints.
   Reveal Solution Discussion Select an answer first
5. 34foundation · easy

   Which of the following is a key benefit of automating data pipeline steps such as transformation and loading?

   • AEliminates the need for data validation
   • BGuarantees zero data loss
   • CReduces manual intervention and human error
   • DRemoves the need for monitoring
   Reveal Solution Discussion Select an answer first
6. 35foundation · easy

   What is the primary role of data management architecture in the context of data pipelines?

   • ATo replace the need for data governance
   • BTo eliminate the need for data pipelines
   • CTo focus only on real-time data processing
   • DTo provide a structured framework for data storage, processing, and integration across the organization
   Reveal Solution Discussion Select an answer first