MySQL按两列分组

Question

I have a table like this

+---------------------+-------+-----+
|       dateTime      | value |  id |
+---------------------+-------+-----+
| 2013-04-06 06:54:00 |  2.5  |  1  |
| 2013-04-06 06:55:00 |  2.9  |  2  |
| 2013-04-06 06:56:00 |  2.4  |  3  |
| 2013-04-06 06:57:00 |  2.6  |  4  |
| 2013-04-06 06:58:00 |  2.5  |  5  |
| 2013-04-06 06:59:00 |  2.6  |  6  |
| 2013-04-06 06:54:00 |  2.8  |  7  |
| 2013-04-06 06:55:00 |  2.5  |  8  |
| 2013-04-06 06:56:00 |  2.1  |  9  |
+---------------------+-------+-----+

I want to apply group by on two columns but query below does not work for me

SELECT * FROM table GROUP BY dateTime, value

I want distinct values from both columns. I have tried this one too but this one is not working either

SELECT distinct(dateTime),value,id FROM table GROUP BY value

I want each column to be unique. My Desired output is

+---------------------+-------+-----+
|       dateTime      | value |  id |
+---------------------+-------+-----+
| 2013-04-06 06:54:00 |  2.5  |  1  |
| 2013-04-06 06:55:00 |  2.9  |  2  |
| 2013-04-06 06:56:00 |  2.4  |  3  |
| 2013-04-06 06:57:00 |  2.6  |  4  |
+---------------------+-------+-----+

Can anyone help me with this? Thanks

Answer 1

If you need to have unique values in both columns (exclude rows with dateTime or value returned in previous rows) try this query.

SELECT a.datetime, a.value, a.id FROM test a
WHERE NOT EXISTS
  (
     SELECT 1 FROM test b 
     WHERE b.id < a.id
       AND (a.datetime = b.datetime OR b.value = a.value)
  )

SQLFiddle

Answer 2

When using grouping predicates, the values selected must be either

a) values used to group the results

b) group functions (like SUM , AVG , COUNT ..)

do something like

SELECT dateTime, value FROM table GROUP BY dateTime, value  (order is not needed to be kept).

if you did something like

SELECT dateTime, value FROM table GROUP BY value

You could end with a table where, to the row with value = 1 , you would have 3 or 4 dateTime values that are different. And that is not a table, so the SQL would be invalid.

Answer 3

If you want both columns to be unique, you can use GROUP or DISTINCT:

DISTINCT

SELECT DISTINCT dateTime, value FROM table

GROUP BY

SELECT dateTime, value FROM table GROUP BY dateTime, value

Answer 4

Your question is lacking clarity, so let me ask it this way.

For any given date/time, that multiple entries exist, you want the value and ID of the first time it occurred. Your sample output does not have any for the :58 and :59 minute entry, so I assume you do not want any having only a single record.

select
      YT2.dateTime,
      YT2.value,
      YT2.ID
   from
      ( select YT.dateTime, min(YT.id) MinID
           From YourTable YT
           group by YT.dateTime
           having count(*) > 1 ) as PreQuery
      JOIN YourTable YT2
         ON PreQuery.MinID = YT2.ID
        AND PreQuery.dateTime = YT2.dateTime

Now, if you wanted it based on the date/time that has the lowest VALUE, the above query would only need to be altered to

  ( select YT.dateTime, min(YT.Value) MinValue

and

     ON PreQuery.MinValue = YT2.Value

Answer 5

You could try this

SELECT dateTime, max(value) FROM table GROUP BY dateTime

This will give you one row for each different dateTime with the maximum value for that group of dateTime. You can use alternatives to max() such as min() or sum().

Without knowing precisely what you are trying to achieve, it is hard to give a more precise answer.