Problem

On a Meta product such as Facebook, you are given raw daily monetization events and need to compute a smoothed revenue-per-user trend. Write a function that returns the 7-day rolling average of daily revenue per user (RPU) for each calendar day.

For each event, you are given a date string in YYYY-MM-DD format, a user_id, and a revenue amount. First aggregate events by day to compute:

• daily revenue = sum of all revenue on that day
• daily active users = number of distinct users with at least one event on that day
• daily RPU = daily revenue / daily active users

Then compute the 7-day rolling average of daily RPU over the current day and the previous 6 days. If fewer than 7 days exist so far, average over all available days.

Return a list of [date, rolling_average] pairs in ascending date order. Round each rolling average to 2 decimal places.

Formal Specification

• Input: events, a list of lists where each item is [date, user_id, revenue]
• Output: a list of [date, rolling_average]

Examples

Example 1

• Input: [["2024-01-01","u1",10],["2024-01-01","u2",20],["2024-01-02","u1",30]]
• Output: [["2024-01-01",15.0],["2024-01-02",22.5]]
• Explanation: Day 1 RPU = 30/2 = 15. Day 2 RPU = 30/1 = 30. Rolling averages are 15.0 and (15 + 30)/2 = 22.5.

Example 2

• Input: [["2024-01-01","u1",5],["2024-01-01","u1",15],["2024-01-02","u2",20]]
• Output: [["2024-01-01",20.0],["2024-01-02",20.0]]
• Explanation: On 2024-01-01, revenue is 20 and distinct users is 1, so RPU = 20.

Constraints

• 1 <= len(events) <= 10^5
• date is a valid YYYY-MM-DD string
• user_id is a non-empty string
• 0 <= revenue <= 10^6