带有月份名称列的间隔统计

Question

上次我使用了间隔，但我认为它不像我想要的那样工作， 我只想声明年份和月份。有哪位大师可以在这里教我吗？

这是作为请求的示例数据：

food_type | sales | date_sales
   pizza     buy     02-01-2015 12:00:00
   pizza     buy     02-01-2015 03:00:00
   pizza     sell     03-01-2015 01:00:00
   pizza     sell     04-01-2015 05:00:00
   pizza     buy     05-01-2015 02:00:00

这是我正在尝试做的示例

DECLARE @Startdate AS datetime = '02-01-2015',
 @Enddate AS datetime = '05-01-2015'


SELECT TYPE_FOOD, COUNT(CASE WHEN STATUS = 'BUY' THEN 1 END) AS [INCOMING], COUNT(CASE WHEN STATUS= 'SELL' THEN 1 END) AS [OUTGOING]
FROM Shop WHERE [Something_Date] BETWEEN @StartDate AND @EndDate

或

SELECT TYPE_FOOD, COUNT(CASE WHEN STATUS = 'BUY' THEN 1 END) AS [INCOMING], COUNT(CASE WHEN STATUS= 'SELL' THEN 1 END) AS [OUTGOING],DATEDIFF(MONTH, DATE_OF_PURCHASE, @YEAR ) AS [YEAR]  
FROM Shop 
GROUP BY TYPE_FOOD, DATEDIFF(MONTH,DATE_OF_PURCHASE, @YEAR )

-- 我不能使用 datediff，因为我想在日期之间..

如果使用datediff，这是我能得到的最接近的查询...不再是一年了...

Food TYPE | MONTH | BUY | SELL 
 pizza       12     10     5       
 pizza       13      3     1       
 ....        ...     ..     ..     
 pizza       32     5      2        

我怎样才能使sql输出像这样：

Food TYPE | MONTH   |   BUY | SELL | 
 pizza       Feb-15     10    5        
 pizza       Mar-15     3     1       
 ....        ...       ..     ..     
 pizza       May-15     5     2   

请高手帮帮我，我还是个sql初学者

Answer 1

我认为您需要按 type_food 进行总和和分组

 select TYPE_FOOD, date_format(date_sales, '%m -  %Y'),
 sum(case when sales = 'buy' then 1 else 0 end) as incoming,
 sum(case when sales = 'sell' then 1 else 0 end) as outgoing
 from shop
 group by type_food