应该为此编写什么查询?

时间:2015-08-07 13:04:43

标签: sql oracle greatest-n-per-group

您好我有银行客户的交易表。表的名称是CUST_TRANSACTION。它有一些字段,如CUST_ID,TRANSACTION_DATE等。我想编写一个查询,以便我可以选择3个特定客户ID的最后一个事务。例如,

TRANSACTION_ID | CUST_ID | TRANSACTION_DATE | TRANSACTION_DESC

100023 | 2 | 15-12-2014 06:00 | Withdraw 1000 dollars
100022 | 5 | 14-12-2014 16:00 | Deposit 1000 dollars 
100021 | 5 | 14-12-2014 15:00 | Withdraw 500 dollars
100020 | 4 | 12-12-2014 15:00 | Withdraw 300 dollars
100019 | 4 | 12-12-2014 14:39 | Deposit 300 dollars
100018 | 2 | 12-12-2014 15:00 | Withdraw 60 dollars

我想获得ids 2,4,5的最后一笔交易

TRANSACTION_ID | CUST_ID | TRANSACTION_DATE | TRANSACTION_DESC

100023 | 2 | 15-12-2014 06:00 | Withdraw 1000 dollars
100022 | 5 | 14-12-2014 16:00 | Deposit 1000 dollars
100020 | 4 | 12-12-2014 15:00 | Withdraw 300 dollars

我应该如何处理它?<​​/ p>

编辑:到目前为止,我已尝试过这个

select * 
from CUST_TRANSACTION 
where CUST_ID IN('2', '5', '4') 
and rownum <4 
order by transaction_date desc 

4 个答案:

答案 0 :(得分:1)

你应该试试这个

SELECT CUST_ID , a.TRANSACTION_ID, TRANSACTION_DATE , TRANSACTION_DESC from CUST_TRANSACTION AS a
INNER JOIN (SELECT MAX(TRANSACTION_ID) TRANSACTION_ID FROM CUST_TRANSACTION GROUP BY CUST_ID) AS b ON a.TRANSACTION_ID=b.TRANSACTION_ID

答案 1 :(得分:1)

使用rank()获取顶部项目,然后仅为每个项目选择排名靠前的行。例如:

SELECT *
FROM (
  select 
    transaction_id,
    cust_id,
    transaction_date,
    transaction_desc,
    rank() OVER (PARTITION BY cust_id ORDER BY cust_id, transaction_date desc) AS rank_id
  from cust_transaction
  ) ct
WHERE ct.rank_id = 1

请参阅SQL Fiddle

答案 2 :(得分:0)

试试这个

select t1.TRANSACTION_ID , 
t1.CUST_ID ,t1.TRANSACTION_DATE ,t1.TRANSACTION_DESC from table t1 
inner join
(
select CUST_ID ,min(TRANSACTION_DATE) as  TRANSACTION_DATE from table
group by CUST_ID
) t2
on t1.CUST_ID=t2.CUST_ID and t1.TRANSACTION_DATE =t2.TRANSACTION_DATE 

答案 3 :(得分:0)

按客户ID分组并获取最大值(transaction_date)。获取具有此客户ID和transaction_date组合的记录