我有一个包含交易的简单表格,我希望每月有多少消费者进行的交易总数超过0且他们的第一笔交易不在那个月。第一笔交易是指客户在该月份第一次购买。
我想要获得的结果如下:
+--------+---------+-----------------------------------+
| Year | Month | NumOfCustomersWithPositiveTotals |
+--------+----------------------------+----------------+
| 2014 | 1 | 22 |
+--------+----------------------------+----------------+
| 2014 | 2 | 10 |
+--------+----------------------------+----------------+
我有一个SQL小提琴,我发现同样的事情,但对于那个月内有第一笔交易的消费者。实际上,我正在寻找的查询是相同的,但对于其他消费者而言。
答案 0 :(得分:2)
我认为这就是你想要的:
SELECT count(consumerId) as NumOfCust, mm, yy FROM
(
SELECT consumerId, month(date) as mm, year(date) as yy, sum(amount) as total, mdate FROM beta.transaction as t
LEFT JOIN (
SELECT min(month(date)) as mdate, consumerId as con FROM beta.transaction
GROUP BY consumerId
) as MinDate ON con = t.consumerId
GROUP BY month(date), consumerId
HAVING mdate < mm AND total > 0
) as res
GROUP BY res.mm;
我将尝试从内到外解释
让我们看看名为minDate的JOIN表有什么:
SELECT min(month(date)) as mdate, consumerId as con FROM beta.transaction
GROUP BY consumerId
-- Here we find the first date of transaction per consumerId
下一步
SELECT consumerId, month(date) as mm, year(date) as yy, sum(amount), mdate FROM beta.transaction as t
LEFT JOIN (
SELECT min(month(date)) as mdate, consumerId as con FROM beta.transaction
GROUP BY consumerId
) as MinDate ON con = t.consumerId
GROUP BY month(date), consumerId
HAVING mdate < mm AND total > 0
-- Here we find total amount per consumerId per month and count only the consumers whose first transact (aka minDate) is lower than current month AND total is greater than 0
最后使用外部SELECT我们计算按月分组的上述结果。 我希望你想要它。
答案 1 :(得分:0)
以下查询会返回正确的结果。答案是基于Akis的答案,在group by
条款中添加了consumerId,并在支票和月份中添加了多年。
select
tyyyy, tmm, count(tcon) as oldkund_real
from
(
select
t.yyyy as tyyyy, t.mm as tmm, t.consumerId as tcon, sum(t.amount) as total, fp.yyyy as fpyyyy, fp.mm as fpmm, fp.consumerId as fpcon
from
(
select
year(date) as yyyy, month(date) as mm, consumerId, amount
from
transaction
) as t
left join
(
select
year(min(date)) as yyyy, month(min(date)) as mm, consumerId
from
transaction
group by
consumerid
) as fp
on
t.consumerId = fp.consumerId
group by
t.yyyy, t.mm, t.consumerId
having
STR_TO_DATE(CONCAT('01,', tmm, ',', tyyyy),'%d,%m,%Y') > STR_TO_DATE(CONCAT('01,', fpmm, ',', fpyyyy),'%d,%m,%Y') and total > 0
) as res
group by
tyyyy, tmm
order by
tyyyy, tmm;
非常感谢Akis!