SQL：如何在列中查找不同的值并显示它们的数量？

Question

我有以下 Oracle SQL 表，其中

product_id

101 和 103 分别在

attr1

和

attr3

列中具有不同的值。

data

| PRODUCT_ID | ATTR1 | ATTR2 | ATTR3 |
|------------|-------|-------|-------|
| 101        | a     | x     | z     |
| 101        | a     | x     | zzz   |
| 101        | aa    | x     | z     |
| 102        | b     | y     | z     |
| 102        | b     | y     | z     |
| 103        | c     | z     | z     |
| 103        | c     | z     | zz    |

我想获得以下输出，其中列出了不同的值及其在列中的计数。

output

| PRODUCT_ID | DESCR            | VALUE_COUNT  |
|------------|------------------|--------------|
| 101        | Issue with attr1 | a(2), aa(1)  |
| 101        | Issue with attr3 | z(2), zzz(1) |
| 103        | Issue with attr3 | z(1), zz(1)  |

我写了一个查询来只获取一列的结果，但是要为实际数据编写它需要相当大的努力，我需要检查 20 多列不同的值。有什么建议可以提高效率吗？

query

WITH data AS (
   
   SELECT 101 product_id, 'a'    attr1, 'x'  attr2, 'z'     attr3    FROM dual UNION ALL
   SELECT 101 product_id, 'a'    attr1, 'x'  attr2, 'zzz'   attr3    FROM dual UNION ALL
   SELECT 101 product_id, 'aa'   attr1, 'x'  attr2, 'z'     attr3    FROM dual UNION ALL
   SELECT 102 product_id, 'b'    attr1, 'y'  attr2, 'z'     attr3    FROM dual UNION ALL
   SELECT 102 product_id, 'b'    attr1, 'y'  attr2, 'z'     attr3    FROM dual UNION ALL
   SELECT 103 product_id, 'c'    attr1, 'z'  attr2, 'z'     attr3    FROM dual UNION ALL
   SELECT 103 product_id, 'c'    attr1, 'z'  attr2, 'zz'    attr3    FROM dual
   
), d1 AS (

   SELECT product_id, 'Issue with attr1' descr 
   FROM data
   GROUP BY product_id 
   HAVING COUNT(DISTINCT attr1) > 1

), d2 AS (
      
   SELECT DISTINCT d1.product_id, d1.descr, data.attr1, COUNT(attr1) OVER (PARTITION BY attr1) cnt
   FROM d1
   INNER JOIN data
      ON d1.product_id = data.product_id
   
)

SELECT product_id, descr, LISTAGG(attr1 || '(' || cnt || ')', ', ') WITHIN GROUP (ORDER BY product_id) value_count
FROM d2
GROUP BY product_id, descr
;

Answer 1

您可以将所有属性反透视到单独的行中，计算每个属性和值的行数，并将其与每个

product_id

的行数进行比较。然后用

listagg

.

聚合错误

这只需要向

inpivot ... for ...

添加更多列。

with prep as (
  select
    sample.*
    /*Rowcount per product_id*/
    , count(1) over(partition by product_id) as rowcnt
  from sample
)
, unp as (
  select
    product_id,
    /*Classify the issue*/
    case
      /*Count per value*/
      when count(1) != rowcnt
      then 'Issue with ' || col
    end as issue,
    /*Count per value*/
    val || '(' || count(1) || ')' as cnt
  from prep
  unpivot(
    val for col in (attr1, attr2, attr3)
  )
  group by
    product_id,
    col,
    val,
    rowcnt
)
/*Aggregate all*/
select
  product_id,
  issue,
  listagg(cnt, ', ') as value_count
from unp
where issue is not null
group by
  product_id,
  issue
order by 1, 2

PRODUCT_ID	问题	VALUE_COUNT
101	ATTR1 问题	a(2), aa(1)
101	ATTR3 问题	z(2), zzz(1)
103	ATTR3 问题	z(1), zz(1)

小提琴

Answer 2

首先尝试 GROUPing，然后围绕它使用 LISTAGG。像这样的东西：

WITH data AS (
   
   SELECT 101 product_id, 'a'    attr1, 'x'  attr2, 'z'  attr3    FROM dual UNION ALL
   SELECT 101 product_id, 'a'    attr1, 'x'  attr2, 'zzz'  attr3    FROM dual UNION ALL
   SELECT 101 product_id, 'aa'   attr1, 'x'  attr2, 'z'  attr3    FROM dual UNION ALL
   SELECT 102 product_id, 'b'    attr1, 'y'  attr2, 'z'  attr3    FROM dual UNION ALL
   SELECT 102 product_id, 'b'    attr1, 'y'  attr2, 'z'  attr3    FROM dual UNION ALL
   SELECT 103 product_id, 'c'    attr1, 'z'  attr2, 'z'  attr3    FROM dual UNION ALL
   SELECT 103 product_id, 'c'    attr1, 'z'  attr2, 'zz' attr3    FROM dual
 
)
SELECT product_id,'Issue with attr1' descr,LISTAGG(value_count,',') WITHIN GROUP (ORDER BY value_count) value_count
  FROM (SELECT product_id,
               attr1||'('||COUNT(*)||')' value_count
          FROM (SELECT product_id,
                       attr1,
                       COUNT(DISTINCT attr1) OVER (PARTITION BY product_id) cnt
                  FROM data)
         WHERE cnt > 1
         GROUP BY product_id,
                  attr1)
 GROUP BY product_id                  
UNION ALL                  
SELECT product_id,'Issue with attr3' descr,LISTAGG(value_count,',') WITHIN GROUP (ORDER BY value_count) value_count
  FROM (                  
        SELECT product_id,
               attr3||'('||COUNT(*)||')' value_count
          FROM (SELECT product_id,
                       attr3,
                       COUNT(DISTINCT attr3) OVER (PARTITION BY product_id) cnt
                  FROM data)
         WHERE cnt > 1
         GROUP BY product_id,
                  attr3)
  GROUP BY product_id

Answer 3

如果你没有太多的属性，你可以这样做：

WITH data AS (
   
   SELECT 101 product_id, 'a'    attr1, 'x'  attr2, 'z'  attr3    FROM dual UNION ALL
   SELECT 101 product_id, 'a'    attr1, 'x'  attr2, 'zz'  attr3    FROM dual UNION ALL
   SELECT 101 product_id, 'aa'   attr1, 'x'  attr2, 'z'  attr3    FROM dual UNION ALL
   SELECT 102 product_id, 'b'    attr1, 'y'  attr2, 'z'  attr3    FROM dual UNION ALL
   SELECT 102 product_id, 'b'    attr1, 'y'  attr2, 'z'  attr3    FROM dual UNION ALL
   SELECT 103 product_id, 'c'    attr1, 'z'  attr2, 'z'  attr3    FROM dual UNION ALL
   SELECT 103 product_id, 'c'    attr1, 'z'  attr2, 'zz' attr3    FROM dual
   
), count_ as (SELECT product_id, COUNT(DISTINCT attr1) attr1
  , COUNT(DISTINCT attr2) attr2
  , COUNT(DISTINCT attr3) attr3
   FROM data
   GROUP BY product_id ),
d2 AS (
      
   SELECT DISTINCT count_.product_id, 'Issue with attr1' descr , data.attr1, COUNT(data.attr1) OVER (PARTITION BY data.attr1) cnt
   FROM count_
   INNER JOIN data
      ON count_.product_id = data.product_id AND count_.attr1 > 1 
   UNION ALL
     SELECT DISTINCT count_.product_id, 'Issue with attr2' descr , data.attr2, COUNT(data.attr2) OVER (PARTITION BY data.attr2) cnt
   FROM count_
   INNER JOIN data
      ON count_.product_id = data.product_id AND count_.attr2 > 1 
  UNION ALL
     SELECT DISTINCT count_.product_id, 'Issue with attr3' descr , data.attr3, COUNT(data.attr3) OVER (PARTITION BY data.attr3) cnt
   FROM count_
   INNER JOIN data
      ON count_.product_id = data.product_id AND count_.attr3 > 1 
)
SELECT product_id, descr, LISTAGG(attr1 || '(' || cnt || ')', ', ') WITHIN GROUP (ORDER BY product_id) value_count
FROM d2
GROUP BY product_id, descr
;

PRODUCT_ID	描述	VALUE_COUNT
101	attr1 问题	a(2), aa(1)
101	attr3 问题	z(3), zz(2)
103	attr3 问题	z(3), zz(2)

小提琴