Gids-gekoppelde quiz · Moeilijk Niveau
Quiz voor AI voor data-analisten
Test door AI opgestelde SQL-risico's, zoals join-fan-out, methoden voor het controleren van AI-analyses en waarom schemacontext en reproduceerbare scripts belangrijk zijn.
Vraag 1 van 8
What was ChatGPT's data analysis feature originally called, and what does it do?
Bekijk elke vraag8 vragen
Antwoorden en uitleg verschijnen terwijl je de quiz doet.
- What was ChatGPT's data analysis feature originally called, and what does it do?
- Excel Copilot, which edits spreadsheets
- Code Interpreter, which writes and runs Python on uploaded files
- Tableau Pulse, which builds dashboards
- SQL Studio, which hosts databases
- What happens when a query joins on a non-unique key?
- All null values are dropped
- Time zone problems are fixed
- Rows are duplicated and totals are inflated
- The query runs faster
- What may a model do when it does not know your schema?
- Invent plausible column or table names
- Refuse to write any SQL
- Always ask a clarifying question first
- Return an empty result
- Which checking method does the guide recommend for AI-generated results?
- Trust the result if the query ran without errors
- Ask the AI whether its answer is correct
- Rerun the query to make it faster
- Reconcile against a known total, such as monthly revenue from the finance report
- What does Simpson's paradox describe?
- Nulls being counted twice
- A time zone mismatch in dates
- A trend in aggregated data reversing direction within subgroups
- Charts exaggerating small differences
- What risk does the guide flag in AI-written narrative reports?
- Ze zijn altijd te kort
- They may state causation where the data only shows correlation
- They refuse to summarize dashboards
- They use too many charts
- Why does the guide prefer asking AI for a cleaning script instead of cleaned data?
- Scripts are always shorter
- AI tools cannot output data
- Pasted data is always larger than a script
- A script can be inspected, rerun and version-controlled, while pasted output may contain silent alterations
- What most improves text-to-SQL accuracy according to the guide?
- Showing the schema: tables, column types, keys, example rows and business definitions
- Writing shorter questions
- Hiding table names for privacy
- Typing the question in uppercase