Question

I’m trying to make a query to get rush hours for everyday on a specific month.
The table I have looks like this:

id      idproduct       created_at
1       021354684       2011-10-01 20:25:48
2       033546835       2011-10-01 20:30:15
3       055965654       2011-10-01 20:45:20
4       012975343       2011-10-02 14:03:36
5       021354684       2011-10-02 15:55:48
6       033546835       2011-10-02 16:30:15
7       055965654       2011-10-02 16:45:20
8       012975343       2011-10-02 18:53:36
9       021354684       2011-10-03 08:55:48
10      033546835       2011-10-03 09:30:15
11      055965654       2011-10-03 14:03:20
12      012975343       2011-10-03 14:03:36

What I try to get is something like this…:

day     rush_hour    number_of_rows
1       20:00        3
2       16:00        5
3       14:00        4

Is it possible to get a table like this? can you guys help me?

I made a mistake, sorry for this. The number of rows should be the total of items sold that day, not in that hour 🙁 sorry.

Answer 1

http://sqlfiddle.com/#!2/5b87b/7

First, count every day’s every hour’s count (into a view, because we will use it twice below):

CREATE VIEW hours AS 
SELECT
  DATE( created_at ) AS d,
  HOUR( created_at ) AS h,
  COUNT(*) AS c
FROM item
GROUP BY DATE(created_at), HOUR(created_at);

Final query:

SELECT
  hours.d AS `day`,
  hours.h AS `rush_hour`,
  hours.c AS `count`

-- get the max count for every day
FROM (
      SELECT
        d,          -- the day 
        MAX(c) as c -- the count
      FROM hours
      GROUP BY d
     ) AS maxc

-- find the actual hour(s) with the max count for every day:
INNER JOIN hours ON hours.c = maxc.c
                AND hours.d = maxc.d;