Data Analysis Quiz
Questions: 16 · 10 minutes
1. A fraud model flags 50 transactions, and 20 of those are actually fraudulent. What is the model's precision for this set of flags?
20%
30%
40%
60%
2. A company wants a report listing every customer, including customers who have never placed an order. Customers is the left table and Orders is the right table. Which join is most suitable?
INNER JOIN
CROSS JOIN
SELF JOIN
LEFT JOIN
3. In a conventional hypothesis test, what does a p-value represent?
Assuming the null hypothesis is true, the probability of results at least as extreme as those observed
The probability that the null hypothesis itself is true
The proportion of observations that must be measurement errors
The probability that another study will produce exactly the same estimate
4. Before splitting data into training and test sets, an analyst calculates a transformation using information from the full dataset. What is the main concern?
The test set will become too large
The model will be unable to make numeric predictions
Information from the test set may leak into model training
The transformation will necessarily remove every outlier
5. A university estimates student commuting habits using a survey distributed only to students living in campus housing. What is the clearest problem?
The sample may not represent students who live off campus
The survey contains too many numeric variables
The survey cannot contain any valid observations
The sample size must equal the entire student population
6. Ice cream sales and sunburn cases both rise during summer. What is the most defensible interpretation?
The association proves that the data were collected incorrectly
Ice cream sales must directly cause sunburn
Sunburn must directly increase ice cream demand
A third factor, such as hot sunny weather, may influence both
7. What is the standard repeated-sampling interpretation of a 95% confidence-interval procedure?
Across repeated samples, about 95% of intervals produced this way would contain the true parameter
The sample statistic equals the population value in 95% of datasets
There is a 95% chance that every observation lies inside one calculated interval
The tested hypothesis has a 95% probability of being correct
8. An analyst wants to compare variables measured in dollars, years, and kilograms on a common standardized scale. Which transformation is designed for this purpose?
Replacing each value with its column total
Sorting every column from smallest to largest
Subtracting the variable mean and dividing by its standard deviation
Converting every positive value to 1
9. A website team wants to estimate whether a new checkout design causes more purchases. Which approach provides the strongest evidence?
Compare this month's purchases with purchases from a year ago
Randomly assign comparable visitors to the old or new design
Ask the design team which version they expect to perform better
Show the new design only to returning customers and compare them with new visitors
10. In a tidy dataset of individual survey responses, what does each row typically represent?
A single variable
A summary statistic
A category label
A single observation
11. You need total sales for each region from a SQL table containing one row per transaction. Which clause is essential for producing separate regional totals?
ORDER BY transaction_id
LIMIT region
GROUP BY region
DISTINCT sales_amount
12. After joining orders to a product-tag table, an analyst notices that some orders appear multiple times because products can have several tags. What could happen if revenue is summed immediately?
Revenue may be overstated because some orders are counted repeatedly
Orders without tags will receive estimated revenue
All duplicated records will be removed by the sum
Revenue will automatically be converted to an average
13. A spreadsheet contains blank values for customer age. Why should an analyst avoid automatically replacing every blank with zero?
Zero values cannot be used in calculations
A missing age is not equivalent to an age of zero
Replacing blanks prevents the data from being sorted
Blank cells are always excluded by statistical software
14. Which chart is generally most appropriate for examining the relationship between advertising spend and sales when both are numeric variables?
A pie chart
A scatter plot
A stacked bar chart
A single-value scorecard
15. A delivery-time model has a mean absolute error of 3.5 minutes. What does this value describe?
The model explains 3.5% of variation in delivery time
Exactly 3.5% of deliveries were predicted incorrectly
The largest prediction error was 3.5 minutes
Predictions differ from actual times by 3.5 minutes on average in absolute terms
16. A small neighborhood has household incomes of $35k, $38k, $41k, $44k, and $900k. Which measure best represents a typical household income there?
The arithmetic mean
The median
The range
The variance