Checkpoint: Event Boundaries and Snapshots

Trace new event histories and diagnose window-query mistakes. Select one option per multiple-choice question; enter only the requested integer for short answers.

Get every question correct to pass this checkpoint. You can review feedback, revise your answers, and submit again.

1. One user has events at 08:00, 08:20, 08:50, 09:21, and 09:51 on the same day. A new session requires a gap greater than 30 minutes. How many sessions are there? Enter one integer.

2. A session query uses LAG ordered by (event_ts, event_id), but its cumulative SUM uses only event_ts with a ROWS frame. What change makes tied events consistent?

Select one answer.

Add ORDER BY user_id to the final result only. Use the same (event_ts, event_id) order in the cumulative window. Remove tied events using DISTINCT event_ts. Group by user_id before applying LAG.

3. At an inclusive noon cutoff, a device has ID 40 at 11:59 (offline), ID 41 at 12:00 (online), ID 43 at 12:00 (error), and ID 44 at 12:01 (online). Which event supplies the snapshot under the course tie policy?

Select one answer.

ID 44: the latest event overall. ID 40: cutoff-time rows are excluded. ID 41: the first event at the latest eligible time. ID 43: the greatest ID at the latest eligible time.

4. A query ranks all device history, then applies rn = 1 AND recorded_at <= cutoff. A device has one event before and one after cutoff. What happens?

Select one answer.

The device disappears because the future event held rank 1. The earlier event is automatically renumbered to rank 1. Both events appear because their timestamps differ. The filter changes the window ordering to ascending.

5. A user has three events at 14:00 with different event IDs and one at 14:30. What event_count belongs to that session? Enter one integer.

6. Which statement about MAX(recorded_at), MAX(status), and MAX(customer_id) grouped per device is correct?

Select one answer.

All three maxima necessarily come from the same event. A final ORDER BY device_id makes the maxima refer to one event. They can combine attributes from different events; select a complete ranked row instead. They are safe whenever event_id is unique.

Auto-advance on pass

Join Discord