Watch 2 video solutions for Hopper Company Queries III, a hard level problem involving Database. This walkthrough by Everyday Data Science has 381 views views. Want to try solving it yourself? Practice on FleetCode or read the detailed text solution.
Table: Drivers
+-------------+---------+ | Column Name | Type | +-------------+---------+ | driver_id | int | | join_date | date | +-------------+---------+ driver_id is the column with unique values for this table. Each row of this table contains the driver's ID and the date they joined the Hopper company.
Table: Rides
+--------------+---------+ | Column Name | Type | +--------------+---------+ | ride_id | int | | user_id | int | | requested_at | date | +--------------+---------+ ride_id is the column with unique values for this table. Each row of this table contains the ID of a ride, the user's ID that requested it, and the day they requested it. There may be some ride requests in this table that were not accepted.
Table: AcceptedRides
+---------------+---------+ | Column Name | Type | +---------------+---------+ | ride_id | int | | driver_id | int | | ride_distance | int | | ride_duration | int | +---------------+---------+ ride_id is the column with unique values for this table. Each row of this table contains some information about an accepted ride. It is guaranteed that each accepted ride exists in the Rides table.
Write a solution to compute the average_ride_distance and average_ride_duration of every 3-month window starting from January - March 2020 to October - December 2020. Round average_ride_distance and average_ride_duration to the nearest two decimal places.
The average_ride_distance is calculated by summing up the total ride_distance values from the three months and dividing it by 3. The average_ride_duration is calculated in a similar way.
Return the result table ordered by month in ascending order, where month is the starting month's number (January is 1, February is 2, etc.).
The result format is in the following example.
Example 1:
Input: Drivers table: +-----------+------------+ | driver_id | join_date | +-----------+------------+ | 10 | 2019-12-10 | | 8 | 2020-1-13 | | 5 | 2020-2-16 | | 7 | 2020-3-8 | | 4 | 2020-5-17 | | 1 | 2020-10-24 | | 6 | 2021-1-5 | +-----------+------------+ Rides table: +---------+---------+--------------+ | ride_id | user_id | requested_at | +---------+---------+--------------+ | 6 | 75 | 2019-12-9 | | 1 | 54 | 2020-2-9 | | 10 | 63 | 2020-3-4 | | 19 | 39 | 2020-4-6 | | 3 | 41 | 2020-6-3 | | 13 | 52 | 2020-6-22 | | 7 | 69 | 2020-7-16 | | 17 | 70 | 2020-8-25 | | 20 | 81 | 2020-11-2 | | 5 | 57 | 2020-11-9 | | 2 | 42 | 2020-12-9 | | 11 | 68 | 2021-1-11 | | 15 | 32 | 2021-1-17 | | 12 | 11 | 2021-1-19 | | 14 | 18 | 2021-1-27 | +---------+---------+--------------+ AcceptedRides table: +---------+-----------+---------------+---------------+ | ride_id | driver_id | ride_distance | ride_duration | +---------+-----------+---------------+---------------+ | 10 | 10 | 63 | 38 | | 13 | 10 | 73 | 96 | | 7 | 8 | 100 | 28 | | 17 | 7 | 119 | 68 | | 20 | 1 | 121 | 92 | | 5 | 7 | 42 | 101 | | 2 | 4 | 6 | 38 | | 11 | 8 | 37 | 43 | | 15 | 8 | 108 | 82 | | 12 | 8 | 38 | 34 | | 14 | 1 | 90 | 74 | +---------+-----------+---------------+---------------+ Output: +-------+-----------------------+-----------------------+ | month | average_ride_distance | average_ride_duration | +-------+-----------------------+-----------------------+ | 1 | 21.00 | 12.67 | | 2 | 21.00 | 12.67 | | 3 | 21.00 | 12.67 | | 4 | 24.33 | 32.00 | | 5 | 57.67 | 41.33 | | 6 | 97.33 | 64.00 | | 7 | 73.00 | 32.00 | | 8 | 39.67 | 22.67 | | 9 | 54.33 | 64.33 | | 10 | 56.33 | 77.00 | +-------+-----------------------+-----------------------+ Explanation: By the end of January --> average_ride_distance = (0+0+63)/3=21, average_ride_duration = (0+0+38)/3=12.67 By the end of February --> average_ride_distance = (0+63+0)/3=21, average_ride_duration = (0+38+0)/3=12.67 By the end of March --> average_ride_distance = (63+0+0)/3=21, average_ride_duration = (38+0+0)/3=12.67 By the end of April --> average_ride_distance = (0+0+73)/3=24.33, average_ride_duration = (0+0+96)/3=32.00 By the end of May --> average_ride_distance = (0+73+100)/3=57.67, average_ride_duration = (0+96+28)/3=41.33 By the end of June --> average_ride_distance = (73+100+119)/3=97.33, average_ride_duration = (96+28+68)/3=64.00 By the end of July --> average_ride_distance = (100+119+0)/3=73.00, average_ride_duration = (28+68+0)/3=32.00 By the end of August --> average_ride_distance = (119+0+0)/3=39.67, average_ride_duration = (68+0+0)/3=22.67 By the end of Septemeber --> average_ride_distance = (0+0+163)/3=54.33, average_ride_duration = (0+0+193)/3=64.33 By the end of October --> average_ride_distance = (0+163+6)/3=56.33, average_ride_duration = (0+193+38)/3=77.00
Problem Overview: You need to report the 3‑month rolling average of ride distance and ride duration for each month of 2020. The result must include all months from January to December, even if no rides happened. Data comes from the rides table joined with accepted ride details, then aggregated by month.
Approach 1: Monthly Aggregation + Window Function (O(n) time, O(n) space)
Start by joining rides with accepted ride information and filtering rows where requested_at falls in 2020. Aggregate rides by month using MONTH(requested_at), computing total distance and duration per month. Since the result must include months with zero rides, generate a fixed month list (1–12) and left join the aggregated data.
Once monthly totals are available, compute the rolling averages using a SQL window function. Use AVG(... ) OVER (ORDER BY month ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) so each row considers the current month and the previous two months. This produces the required 3‑month moving average directly in SQL. Window functions make this clean and efficient because they avoid repeated scans or correlated subqueries. Time complexity is O(n) for scanning and aggregating rides, with O(n) intermediate storage for monthly results.
This pattern appears frequently in analytics queries where you need moving averages or running totals. Window functions are part of advanced SQL and are heavily used in database reporting workloads. Understanding the window functions syntax makes problems like this straightforward.
Approach 2: Self Join for Rolling Window (O(n^2) time, O(n) space)
Another way to compute the 3‑month average is by first generating monthly aggregates and then joining each month with its previous two months. For every month m, join rows where month BETWEEN m-2 AND m and compute the average distance and duration across those rows. This works even without window function support.
The drawback is performance and complexity. Each month potentially scans multiple rows, producing roughly O(n^2) work in the worst case. The SQL also becomes harder to read compared with the concise window frame syntax.
Recommended for interviews: The window function approach is the expected solution. It shows you understand SQL analytics features and can compute rolling metrics efficiently. The self‑join version demonstrates the concept but lacks the elegance and performance of the window function solution.
| Approach | Time | Space | When to Use |
|---|---|---|---|
| Monthly Aggregation + Window Function | O(n) | O(n) | Best approach when the database supports SQL window functions. Clean and efficient for rolling averages. |
| Self Join Rolling Window | O(n^2) | O(n) | Useful in older SQL engines without window functions, but less efficient and harder to maintain. |
