选择低于另一个值的值-Oracle

Question

我有一个包含各种信息的表，我需要选择以下值：

1）的cod_anag_prov = 0或= 2
2）计数（1）> 1

然后为所有涉及点1和2且所有事件中具有最小count（1）的单个记录将标志设置为1。
我想到了使用density_rank函数，并弥补了这一点：

SELECT PDRA,
    COD_DISTRIBUTORE_STARGAS,
    ANNOMESE_RIF,
   MAX_CNT,
    MIN(MAX_CNT) KEEP (DENSE_RANK FIRST ORDER BY MAX_CNT) OVER (PARTITION BY PDRA) OCCORRENZA_MINORE
    FROM 
    (SELECT PDRA,
    COD_DISTRIBUTORE_STARGAS,
    ANNOMESE_RIF,
    COUNT(1) AS MAX_CNT
  FROM STM_VOLUME_AGGR
 WHERE (COD_ANAGR_PROV = 0
  OR COD_ANAGR_PROV    = 2)
   GROUP BY PDRA,
    COD_DISTRIBUTORE_STARGAS,
    ANNOMESE_RIF
 HAVING COUNT(1)>1
  ORDER BY PDRA);

到目前为止（我认为），我已经能够做到我之前说的话。
现在，如果我得到以下结果：

    34624200    1905    201305  6   6
    34624200    83      201305  13  6
    34624200    93      201305  14  6
    34439201    1       201305  11  2
    34439201    6       201305  2   2

我想将行的标志设置为1：

 34624200   1905    201305  6   6
 34439201   6       201305  2   2

我该怎么办？
我知道我所做的事情要复杂得多，但现在我的大脑正在融化xD（我对SQL还是很陌生）...


更新1：好的，我已经完成了，但是肯定需要优化它。 费用为3.300.000：S

这是我的解决方案：

SELECT PDRA,
    COD_DISTRIBUTORE_STARGAS,
    ANNOMESE_RIF,
   MIN(MAX_CNT),
    NUMERO_OCCORRENZE FROM
 (SELECT PDRA,
    COD_DISTRIBUTORE_STARGAS,
    ANNOMESE_RIF,
   MAX_CNT,
    MIN(MAX_CNT) KEEP (DENSE_RANK FIRST ORDER BY MAX_CNT) OVER (PARTITION BY PDRA) NUMERO_OCCORRENZE
    FROM
    (SELECT PDRA,
    COD_DISTRIBUTORE_STARGAS,
    ANNOMESE_RIF,
    COUNT(1) AS MAX_CNT
  FROM STM_VOLUME_AGGR
 WHERE (COD_ANAGR_PROV = 0
  OR COD_ANAGR_PROV    = 2)
   GROUP BY PDRA,
    COD_DISTRIBUTORE_STARGAS,
    ANNOMESE_RIF
 HAVING COUNT(1)>1
  ORDER BY PDRA))
  GROUP BY 
  PDRA, COD_DISTRIBUTORE_STARGAS, ANNOMESE_RIF, NUMERO_OCCORRENZE
  HAVING MIN(MAX_CNT)=NUMERO_OCCORRENZE
  ;

显然，我不确定这是否是最好的解决方案（即使它可以工作）...

Answer 1

您的查询需要简化，但是快速取胜是；

    SELECT PDRA,
    COD_DISTRIBUTORE_STARGAS,
    ANNOMESE_RIF,
   MAX_CNT,
   row_number() over(partititon by PDRA order by MAX_CNT) rank_id
    FROM 
    (SELECT PDRA,
    COD_DISTRIBUTORE_STARGAS,
    ANNOMESE_RIF,
    COUNT(1) AS MAX_CNT
  FROM STM_VOLUME_AGGR
 WHERE (COD_ANAGR_PROV = 0
  OR COD_ANAGR_PROV    = 2)
   GROUP BY PDRA,
    COD_DISTRIBUTORE_STARGAS,
    ANNOMESE_RIF
 HAVING COUNT(1)>1
  ORDER BY PDRA)

Answer 2

如果我理解这种情况，当OCCORRENZA_MINORE = MAX_CNT时：

UPDATE STM_VOLUME_AGGR
   SET flag = 1
 WHERE (PDRA, COD_DISTRIBUTORE_STARGAS, ANNOMESE_RIF) IN 
       (SELECT PDRA, COD_DISTRIBUTORE_STARGAS, ANNOMESE_RIF
          FROM ( your_query )
         WHERE OCCORRENZA_MINORE = MAX_CNT)
   AND (COD_ANAGR_PROV = 0 OR COD_ANAGR_PROV = 2)

Answer 3

如前所述是group by，也可以是另一列，您可以通过“ case”和查询的最后两列之间的比较进行填充（case ...等于1，否则0结束）。