Totals that keep every row

Computing · Data & Databases · ages 19-20

Name ______________________ Date ____________

What does PARTITION BY do?
- Deletes duplicate rows
- Restarts the math per group
- Sorts the whole table forever
What survives a window function?
- Every detail row
- Only one row per group
- Only the column names
GROUP BY keeps every detail row in the output.
Circle one: True False
Two orders tie for second. Which rank form leaves a gap?
- ROW_NUMBER
- DENSE_RANK
- RANK
Which query shows each order beside its customer running total?
- SELECT customer, date, amount, SUM(amount) OVER (PARTITION BY customer ORDER BY date) FROM Orders
- SELECT customer, SUM(amount) FROM Orders GROUP BY customer
- SELECT DISTINCT customer FROM Orders
Why can GROUP BY not give the same output as the running total?
- It runs too fast
- It collapses the rows the totals sit beside
- It forbids all sums
A self join is always clearer than a window for running totals.
Circle one: True False
A report needs ties sharing rank 2 with no rank skipped. Which form?
- RANK
- ROW_NUMBER
- DENSE_RANK

Answer key

For grown-ups. Fold this page away before handing over the rest.

Restarts the math per group · Each partition gets a fresh calculation.
Every detail row · Windows calculate without collapsing rows.
False · Grouping collapses each group to one row.
RANK · RANK shares a rank then skips the next.
SELECT customer, date, amount, SUM(amount) OVER (PARTITION BY customer ORDER BY date) FROM Orders · Only the window keeps rows beside totals.
It collapses the rows the totals sit beside · Collapsed rows cannot carry per-row totals.
False · The window states partition and order in one pass.
DENSE_RANK · DENSE_RANK shares ranks and packs tight.