SQL MIN（值）匹配日期

Question

Alright I have a table where I need to get the relevant date from a GROUP BY function...

Here is a sample of the data:
min_price, day
140000     2010-09-26
130000     2010-09-27
154991     2010-10-08
143000     2010-10-09
156470     2010-10-10

I would like this result:
130000     2010-09-27
143000     2010-10-09

Here is the SQL I use:

SELECT MIN(min_price) AS low_price, min_price, day
FROM compress
WHERE item_name = '$item'
GROUP BY EXTRACT(MONTH FROM day), EXTRACT(YEAR FROM day)

Here is what I get:
130000     2010-09-26
143000     2010-10-08

As you can see, the dates don't match up with the min row value. I need them to match up. What SQL can I use to get that result?

Answer 1

Select C.min_price, C.day
From compress As C
Where C.item_name = '$item'
    And Exists  (
                Select 1
                From compress As C2
                Where C2.item_name = C.item_name
                    And Extract(Month From C2.day) = Extract(Month From C.day)
                    And Extract(Year From C2.day) = Extract(Year From C.day)
                Group By Extract(Month From C2.day), Extract(Year From C2.day)
                Having C.min_price = Min(C2.min_price)
                )

ADDITION

MySQL is better with joins so here's a variant of the same query

Select C.min_price, C.day
From compress As C
    Join    (
            Select C2.item_name
                , Extract(Month From C2.day) As ItemMonth
                , Extract(Year From C2.day) As ItemYear
                , Min(C2.min_price) As MinPrice
            From compress As C2
            Group By C2.item_num, Extract(Month From C2.day), Extract(Year From C2.day)
            ) As Z
        On Z.item_name = C.item_name
            And Z.ItemMonth = Extract(Month From C.day)
            And Z.ItemYear = Extract(Year From C.day)
            And Z.MinPrice = C.min_price
Where C.item_name = '$item'

What is really slowing down any variation is the calling of Extract on every row. If you had a calendar table containing a row for each date and a column for the month and year of the date and your calendar table covered the date ranges in your data, you could then do something like:

Select C.min_price, C.day
From compress As C
    Join Calendar As Cal
        On Cal.Date = C.Day
    Join    (
            Select C3.item_name
                , Cal3.Month, Cal3.Year
                , Min(C3.min_price) As MinPrice
            From compress As C3
                Join Calendar As Cal3
                    On Cal3.Date = C3.day
            Group By C3.item_num, Cal3.Month, Cal3.Year
            ) As Z
        On Z.item_name = C.Item_name
            An Z.Month = Cal.Month
            And Z.Year = Cal.Year
            And Z.MinPrice = C.min_price
Where C.item_name = '$item'

Answer 2

SELECT min_price AS low_price, min_price, day
FROM compress
WHERE item_name = '$item' AND min_price = MIN(min_price)
GROUP BY EXTRACT(MONTH FROM day), EXTRACT(YEAR FROM day)

This will then select the row where the min_price is the same as the MIN(min_price)

Answer 3

Your SQL is not valid. When grouping, every column in the display list (that is not part of the group by expression) must be passed through an aggregate function, eg AVG(), MIN(), MAX() but you're just asking for "day".

Please state in natural language (English) what you are trying to achieve. Your example is not perfectly clear.

Answer 4

If I have understood the problem correctly, this should work...

mysql> select * from stack;
+-----------+------------+
| min_price | myday      |
+-----------+------------+
|    140000 | 2010-09-26 | 
|    130000 | 2010-09-27 | 
|    154991 | 2010-10-08 | 
|    143000 | 2010-10-09 | 
|    156470 | 2010-10-10 | 
+-----------+------------+
5 rows in set (0.00 sec)

mysql> select a.* from stack as a inner join 
    -> (select min(min_price) as my_price from stack group by extract(YEAR_MONTH from myday)) as b
    -> on a.min_price = b.my_price;
+-----------+------------+
| min_price | myday      |
+-----------+------------+
|    130000 | 2010-09-27 | 
|    143000 | 2010-10-09 | 
+-----------+------------+
2 rows in set (0.00 sec)

You will need to join on a primary key in order to avoid unexpected results. If the query is slow, create a temporary table for subselect, build indexes on it and then use the table to join with the main table.