无法从sql查询中检索所需的数据

时间:2012-01-09 12:49:48

标签: sql oracle oracle10g analytic-functions

SELECT     
     a.alloc_date,
     p.plan,
     p.emp_id,
      a.veh,
     a.contri_type,
     a.amount,
     SUM (a.alloc_qty) AS sum_alloc_qty,                    -- 1000 funds distributed
     SUM (a.alloc_qty * a.amount) AS sum_alloc_value,       -- 1000*2 = 2000 
     COUNT (DISTINCT part_id) AS sum_emp_count,             -- 4 employees 
     MAX (a.alloc_qty * a.amount) AS max_value_to_one_emp,  --  600 
     null as "emp_count_with_max_value"                     --  Unable to retrieve -        idealy answer should be 3 in this example
 FROM   
     alloc a, emp p
 WHERE   
     A.alloc_date >= TO_DATE ('20111001', 'YYYYMMDD')
     AND a.emp_id = p.emp_id
GROUP BY   
     a.alloc_date,
     p.plan,
     p.emp_id,
     a.veh,
     a.contri_type,
     a.amount
ORDER BY   
     alloc_date, emp_id, amount

此处,现有查询的工​​作方式如下。

假设,该公司正在分配1000只基金,其中每只基金的价格为2。

现在这1000个基金分配给4名员工。

基本问题是检索一名员工的最大资金价值。假设,资金分配是:

emp1=600 (300*2), emp2=600 (300*2), emp3=600 (300*2), emp4=300 (300*1)

所以,这里一个雇员的资金最大值= 600。

我现在可以通过查询检索。

但是,现在下一个问题是检索另一列(emp_count_with_max_value),该列应该具有在每个组下获得此最大值的员工数量。

在我们的例子中,结果是3名员工。但我无法检索相同的

这里我只给出了一组的数据。结果查询输出应如下所示:

'11/12/86','abc','E25','pqr','qvr',2,1000,2000,4,600,3

2 个答案:

答案 0 :(得分:1)

基本上,如果您对员工的分配进行排名,很容易确定谁获得了最大的金额。然后你需要有一个外部查询来计算幸运狗的数量。

select alloc_date,
       plan,
       emp_id,
       veh,
       contri_type,
       amount,
       sum_alloc_qty,                    
       sum_alloc_value,      
       sum_emp_count,      
       max_value_to_one_emp,  
       sum ( case when rnk = 1 then 1 else 0 end ) as emp_count_with_max_value
from (
    SELECT     a.alloc_date,
           p.plan,
           p.emp_id,
           a.veh,
           a.contri_type,
           a.amount,
           SUM (a.alloc_qty) AS sum_alloc_qty,                    -- 1000 funds distributed
           SUM (a.alloc_qty * a.amount) AS sum_alloc_value,       -- 1000*2 = 2000 
           COUNT (DISTINCT part_id) AS sum_emp_count,             -- 4 employees 
           MAX (a.alloc_qty * a.amount) AS max_value_to_one_emp,  --  600 
           dense_rank() over  (order by a.alloc_qty desc)      rnk  -- rank allocation in descending order
           FROM   alloc a, emp p
       WHERE   A.alloc_date >= TO_DATE ('20111001', 'YYYYMMDD')
           AND a.emp_id = p.emp_id
    GROUP BY   a.alloc_date,
           p.plan,
           p.emp_id,
           a.veh,
           a.contri_type,
           a.amount
)
group by alloc_date,
       plan,
       emp_id,
       veh,
       contri_type,
       amount,
       sum_alloc_qty,                    
       sum_alloc_value,      
       sum_emp_count,      
       max_value_to_one_emp
ORDER BY   alloc_date,
       emp_id,
       amount

注意:在没有测试数据的情况下,我没有测试过这段代码,我也不保证它没有错误。但原则是合理的。

答案 1 :(得分:1)

你也必须通过alloc_qty进行分组,但是如果你使用它:

count(*) keep (dense_rank last order by A.ALLOC_QTY * A.AMOUNT) over () as "emp_count_with_max_value"

那应该返回与最大值匹配的记录数。因此,集成到您的代码中应该是:

SELECT     a.alloc_date,
       p.plan,
       p.emp_id,
       a.veh,
       a.contri_type,
       a.amount,
       SUM (a.alloc_qty) AS sum_alloc_qty,                    -- 1000 funds distributed
       SUM (a.alloc_qty * a.amount) AS sum_alloc_value,       -- 1000*2 = 2000 
       COUNT (DISTINCT part_id) AS sum_emp_count,             -- 4 employees 
       MAX (a.alloc_qty * a.amount) AS max_value_to_one_emp,  --  600 
       count(*) keep (dense_rank last order by A.ALLOC_QTY * A.AMOUNT) over () as "emp_count_with_max_value"                     --  Unable to retrieve -           idealy answer should be 3 in this example
    FROM   alloc a, emp p
   WHERE   A.alloc_date >= TO_DATE ('20111001', 'YYYYMMDD')
       AND a.emp_id = p.emp_id
GROUP BY   a.alloc_qty,
       a.alloc_date,
       p.plan,
       p.emp_id,
       a.veh,
       a.contri_type,
       a.amount
ORDER BY   alloc_date,
       emp_id,
       amount

您可能还想将sum_emp_count更改为

COUNT (DISTINCT part_id) over ()

和max_value_to_one_emp到

MAX (a.alloc_qty * a.amount) over ()

否则,您将无法获得所有数据集的值。