Викторина, связанная с руководством · Средний Уровень
Приглашение к тесту по анализу данных с помощью ИИ
Проверяется контекст набора данных, контрольные точки метода, варианты файловых инструментов, проверка вычислений, проверки соединения и отсутствия данных, а также конфиденциальность.
Вопрос 1 из 8
What information helps an assistant interpret dataset values correctly?
Предварительный просмотр каждого вопроса8 вопросы
Ответы и пояснения появляются по мере прохождения теста.
- What information helps an assistant interpret dataset values correctly?
- Column names and units, without describing how or when values were collected
- Column meanings, units, time range, collection context, and limitations
- A summary of the desired chart, without clarifying what each field represents
- An instruction to infer undocumented definitions from correlations
- Why ask the assistant to state its proposed analysis method before interpreting results?
- It lets the assistant select an analysis method before seeing any data
- It proves the selected method will produce an unbiased estimate
- It automatically resolves ambiguous field definitions from the dataset
- It creates a checkpoint where a mistaken assumption can be corrected
- What does OpenAI’s data-analysis documentation describe for ChatGPT?
- File upload, interactive table views, natural-language questions, comparisons, and charts or analyses
- File upload and natural-language questions, with all workspaces sharing identical capabilities
- Interactive tables and charts that require no review of source data
- Comparisons and analyses that are guaranteed to use the correct statistical method
- A conversion rate is defined as completed purchases divided by eligible checkout sessions. Which denominator should the calculation use?
- The count of distinct products viewed during the period
- The number of completed purchases, which is the numerator
- The count of eligible checkout sessions in the same analysis period
- All site visits, including visitors who never entered checkout
- An order-lines table has many rows per order, while a product table should have one row per SKU. What check helps prevent a join from multiplying rows?
- Aggregate order lines to one row per order before joining to products
- Join on product name and remove repeated output rows afterward
- Check that the tables have the same number of rows before joining
- Verify the product table has one matching row per SKU and inspect the join cardinality
- A model reports monthly revenue. Which method provides the strongest independent check of the total?
- Re-run the same assistant-generated code without checking its inputs or filters
- Recalculate revenue from source rows with a separate reviewed formula or script and the same documented filters
- Ask the same model to repeat its total from the same computed table
- Compare only the total with a rounded chart label from the same analysis
- Before calculating a monthly mean with blank measurements, what should the analyst do?
- Inspect why values are missing and choose an appropriate treatment for that field
- Carry the previous value forward for all fields without checking their meaning
- Drop every row with any blank, regardless of how much data that removes
- Replace every blank with zero so each month has the same row count
- What privacy practice does the guide recommend before uploading data?
- Use an approved tool while including every available personal field
- Assume a private workspace waives the organization’s data-handling rules
- Use only approved tools and workspaces and remove unnecessary identifiers
- Remove direct identifiers but upload through an unapproved public service