Question

Consider the following table (“tweets”):

+---------+------------+----------+
| user_id | _date      | tweet_id |
+---------+------------+----------+
| 1       | 2011-03-17 | 12       |
| 1       | 2011-03-17 | 16       |
| 1       | 2011-01-18 | 17       |
| 2       | 2011-03-18 | 23       |
| 2       | 2011-01-17 | 29       |
| 2       | 2011-03-17 | 32       |
| 2       | 2011-01-17 | 33       |
| 2       | 2011-03-18 | 34       |
| 2       | 2011-01-19 | 45       |
| 2       | 2011-03-19 | 48       |
| 3       | 2011-01-19 | 50       |
| 3       | 2011-03-19 | 51       |
| 3       | 2011-01-19 | 52       |
| 3       | 2011-01-20 | 53       |
+---------+------------+----------+

I want the following output: tweets PER user ORDERED BY date:

user_id  _date       cnt
------------------------
1        2011-03-17  2
1        2011-03-18  1
2        2011-03-17  3
2        2011-03-18  2
2        2011-03-19  2
3        2011-03-19  3
3        2011-03-20  1

I don’t understand why, but this doesn’t work:

SELECT user_id, _date, COUNT(tweet_id) AS cnt FROM tweets 
GROUP BY nyse_date ORDER BY sp100_id ASC

Could someone explain to me why?

Answer 1

You almost got it. Not sure why you have nyse_date and sp100_id in there? You need to group together user_id and _date, and order by them. This should do it for you:

SELECT user_id, _date, COUNT(tweet_id) AS cnt FROM tweets 
GROUP BY user_id, _date ORDER BY user_id, _date ASC