New Users Daily Count

Write an SQL query that reports for every date within at most 90 days from today, the number of users that logged in for the first time on that date. Assume today is 2019-06-30.

Example:

Table: Traffic:
+---------+----------+---------------+
| user_id | activity | activity_date |
+---------+----------+---------------+
| 1       | login    | 2019-05-01    |
| 1       | homepage | 2019-05-01    |
| 1       | logout   | 2019-05-01    |
| 2       | login    | 2019-06-21    |
| 2       | logout   | 2019-06-21    |
| 3       | login    | 2019-01-01    |
| 3       | jobs     | 2019-01-01    |
| 3       | logout   | 2019-01-01    |
| 4       | login    | 2019-06-21    |
| 4       | groups   | 2019-06-21    |
| 4       | logout   | 2019-06-21    |
| 5       | login    | 2019-03-01    |
| 5       | logout   | 2019-03-01    |
| 5       | login    | 2019-06-21    |
| 5       | logout   | 2019-06-21    |
+---------+----------+---------------+
Result:
+------------+-------------+
| login_date | user_count  |
+------------+-------------+
| 2019-05-01 | 1           |
| 2019-06-21 | 2           |
+------------+-------------+

Approach

Mysql: Using DATEDIFF method

SELECT login_date, COUNT(user_id) user_count
FROM 
    (SELECT user_id, MIN(activity_date) as login_date
    FROM Traffic
    WHERE activity='login'
    GROUP BY user_id) as t
WHERE DATEDIFF('2019-06-30', login_date) <= 90
GROUP BY login_date

Mysql: Using the DATE_ADD method

SELECT login_date, count(user_id) AS user_count
FROM (
    SELECT user_id, MIN(activity_date) AS login_date
    FROM Traffic
    WHERE activity = 'login'
    GROUP BY user_id) AS t
WHERE login_date >= DATE_ADD('2019-06-30', INTERVAL -90 DAY) 
    AND login_date <= '2019-06-30'
GROUP BY login_date;