[英]wrong count on the multiple joins with the same table
i have 2 tables 我有2张桌子
1 - coupons 1-优惠券
2 - tractions 2-牵引力
for each coupon there might be couple of rows in tractions table I want to have list of all coupons and count of its tractions under different condition 对于每个优惠券,牵引力表中可能会有几行,我想拥有所有优惠券的列表以及在不同条件下的牵引力计数
SELECT `coupons`.`id` ,
count( tractions_all.id ) AS `all` ,
count( tractions_void.id ) AS void,
count( tractions_returny.id ) AS returny,
count( tractions_burned.id ) AS burned
FROM `coupons`
LEFT JOIN `tractions` AS `tractions_all`
ON `coupons`.`id` = `tractions_all`.`coupon_parent`
LEFT JOIN `tractions` AS `tractions_void`
ON `coupons`.`id` = `tractions_void`.`coupon_parent`
AND `tractions_void`.`expired` =1
LEFT JOIN `tractions` `tractions_returny`
ON `tractions_returny`.`coupon_parent` = `coupons`.`id`
AND `tractions_returny`.`expired` =11
LEFT JOIN `tractions` `tractions_burned`
ON `tractions_burned`.`coupon_parent` = `coupons`.`id`
AND `tractions_burned`.`expired` =0
AND '2014-02-12'
WHERE `coupons`.`parent` =0
GROUP BY `coupons`.`id`
right now only one of my coupons has 2
traction on both are burned traction
other coupons have no tractions at all 现在,我的一张优惠券上只有
2
牵引力都burned traction
其他的优惠券根本没有牵引力
here is the result 这是结果
as you can see coupon with id=13
has 4 traction while it should be 2 ... what am i doing wrong ? 如您所见,
id=13
优惠券的牵引力为4,而应该为2 ...我在做什么错? if i remove the last join it works fine and i get 2 如果我删除最后一个连接,它工作正常,我得到2
You are aggregating along multiple dimensions at one time, resulting in a cartesian product for each id. 您一次沿多个维度进行汇总,因此每个ID都会产生笛卡尔积。
If your data volume is not very large, the easiest way to fix this is using distinct
: 如果您的数据量不是很大,则解决此问题的最简单方法是使用
distinct
:
SELECT `coupons`.`id` ,
count(distinct tractions_all.id ) AS `all` ,
count(distinct tractions_void.id ) AS void,
count(distinct tractions_returny.id ) AS returny,
count(distinct tractions_burned.id ) AS burned
If your data is large, then you will probably need to aggregate values as subqueries first and then do the joins. 如果数据很大,则可能需要先将值聚合为子查询,然后再进行联接。
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.