Question

以下是我正在尝试的源数据获取最大计数以及相应的hour_id和id值。对于某个日期应该只有一行，现在有特定日期的多个值。

   id            hour_id    count   date
    621f50772a36e7  23      14474   20141202
    621f50157c2973  0       7190    20141203
    621f5077582f54  7       5043    20141225
    621f505247c107  11      5023    20141224
    621f50251c8b33  10      4943    20141224
    621f5076c9327b  18      4901    20150113
    621f50044c300e  10      4868    20141212
    621f500e10fa5d  12      4858    20141224
    621f505242ec27  9       4843    20141224
    621f505263bc56  14      4716    20141231
    621f50774a456c  19      4712    20141206
    621f5077414404  19      4674    20141123
    621f5077362f46  19      4666    20141224
    621f505246ea97  10      4662    20141225
    621f50522c6bf5  13      4626    20141226
    621f5076c87607  13      4586    20141231
    621f5052297007  17      4574    20141224

我试过这个：

select max(count) cc
,partition_date
,location_id
,hour_id 
from 
(
     select 
     l.location_id
     , substr(x.evt_timestamp,9,2) as hour_id
     , count(1) as count 
     ,partition_date
     from prismeventdetails x
     join l_cellsite_location l
     on x.evt_location = l.location_id
     where x.evt_type = '100'           
     group by l.location_id
     ,hour_id 
     ,partition_date
     order by  cc desc  limit 500
) c 
group by partition_date,location_id,hour_id

但无法获得所需的结果。为一个日期获取多行。因为我已将hour_id列入分组。我想要hour_id和id值，其中count是该日期的最大值。需要帮助，早期回应将不胜感激。

Answer 1

以下是GROUPed和MAXed表与原始表的连接，应该是您想要的。

WITH maxed AS
(
  SELECT max(count) AS max_count, date_
  FROM tests.so_30595512
  GROUP BY date_
)

SELECT maxed.date_, maxed.max_count, t2.hour, t2.id
FROM maxed
JOIN tests.so_30595512 AS t2
ON maxed.date_ = t2.date_
AND maxed.max_count = t2.count;

这是您在上面列出的表格上的一个查询，但同样的方法可以轻松转换为您在示例查询中使用的任何联接表格。

Answer 2

尝试在下面将MAX移动到子查询：

select counts cc
,partition_date
,location_id
,hour_id 
from 
(
     select 
     l.location_id
     , substr(x.evt_timestamp,9,2) as hour_id
     , max(count(1)) as counts 
     ,partition_date
     from prismeventdetails x
     join l_cellsite_location l
     on x.evt_location = l.location_id
     where x.evt_type = '100'           
     group by partition_date, location_id, hour_id
     order by  cc desc  limit 500
)c

Answer 3

try this one:

select [count] cc
,partition_date
,location_id
,hour_id 
from 
(
     select 
     l.location_id
     , substr(x.evt_timestamp,9,2) as hour_id
     , count(1) as [count] 
     ,partition_date
     from prismeventdetails x
     join l_cellsite_location l
     on x.evt_location = l.location_id
     where x.evt_type = '100'           
     group by l.location_id
     ,hour_id 
     ,partition_date
,ROW_NUMBER() over (ORDER BY [count] DESC) AS Number
     order by  cc desc  limit 500
) c 
WHERE Number = 1
group by partition_date,location_id,hour_id

如何获取日期的最大计数和相应的hour_id？

3 个答案: