Question

我有一套信用卡交易的MySQL数据集。

create table trans (
  card_id int,
  amount int
);

insert into trans values (1, 1);
insert into trans values (2, 1);
insert into trans values (3, 1);
insert into trans values (4, 1);
insert into trans values (5, 1);
insert into trans values (5, 1);
insert into trans values (6, 1);
insert into trans values (6, 1);
insert into trans values (7, 1);
insert into trans values (7, 1);
insert into trans values (8, 1);
insert into trans values (8, 1);
insert into trans values (8, 1);
insert into trans values (9, 1);
insert into trans values (9, 1);
insert into trans values (9, 1);
insert into trans values (10, 1);
insert into trans values (10, 1);
insert into trans values (10, 1);
insert into trans values (10, 1);

我想知道：

1. how many cards were used to make at least 1 transaction
2. how many cards were used to make at least 5 transactions
3. how many cards were used to make at least 10 transactions
4. how many cards were used to make at least 20 transactions
etc...

因为组重叠，所以看起来条件聚合是一种更好的方法：

select sum(cnt >= 1) as trans_1,
       sum(cnt >= 5) as trans_5,
       sum(cnt >= 10) as trans_10,
       sum(cnt >= 20) as trans_20
from (select card_id, count(*) as cnt
      from trans
      group by card_id 
      ) d;

问题是上面会在列中生成结果集，但我试图在行中生成结果集。

MySQL将列转换为行的传统方式是重复使用表联合中的select sum（）序列，但在这种情况下，基本数据是派生表，因此该方法似乎不起作用。有关如何将列翻转到行的任何想法吗？

http://sqlfiddle.com/#!9/0f741/3

Answer 1

您可以通过在计算后旋转行或在单独的行上进行计算来执行此操作。第一个应该有更好的表现：

select x.which,
       (case when x.n = 1 then trans_1
             when x.n = 2 then trans_5
             when x.n = 3 then trans_10
             when x.n = 4 then trans_20
        end) as numtransactions
from (select sum(cnt >= 1) as trans_1,
             sum(cnt >= 5) as trans_5,
             sum(cnt >= 10) as trans_10,
             sum(cnt >= 20) as trans_20
      from (select card_id, count(*) as cnt
            from trans
            group by card_id 
           ) d
     ) d join
     (select 1 as n, '1 or more' as which union all
      select 2, '5 or more' union all
      select 3, '10 or more' union all
      select 4, '20 or more'
     ) x
order by x.n;

Answer 2

这是一个简单的答案，但它确实有效。它的优化程度稍差，因为查询需要多次执行才能进行分组和计数。

select sum(cnt >= 1) as trans_1
from (select card_id, count(*) as cnt
      from trans
      group by card_id 
      ) d
UNION ALL
select sum(cnt >= 2) as trans_2
from (select card_id, count(*) as cnt
      from trans
      group by card_id 
      ) d
UNION ALL
select sum(cnt >= 3) as trans_10
from (select card_id, count(*) as cnt
      from trans
      group by card_id 
      ) d
UNION ALL
select sum(cnt >= 4) as trans_20
from (select card_id, count(*) as cnt
      from trans
      group by card_id 
      ) d;

SQL如何计算行中至少有1,5,10,20个等交易的信用卡数量

2 个答案: