如何在 MySQL 中查找重复记录，但有一定程度的差异？

Question

假设我有以下表结构和数据：

+------------------+-------------------------+--------+
| transaction_date | transaction_description | amount |
+------------------+-------------------------+--------+
| 2020-08-20       | Burger King             |  10.06 |
| 2020-08-23       | Burger King             |  10.06 |
| 2020-08-29       | McDonalds               |   6.48 |
| 2020-09-04       | Wendy's                 |   7.45 |
| 2020-09-05       | Dairy Queen             |  14.36 |
| 2020-09-06       | Wendy's                 |   7.45 |
| 2020-09-13       | Burger King             |  10.06 |
+------------------+-------------------------+--------+

我希望能够找到描述和金额匹配的重复交易，但日期之间会有一定程度的差异 +/- 3 天。

因为“汉堡王”的交易是在三天内（2020-08-20 和 2020-08-23），所以它们会被计算为重复，但 2020-09-13 的条目不会.

到目前为止，我有以下查询，但是让我感到困惑的是方差程度。

SELECT t.transaction_date, t.transaction_description, t.amount
FROM transactions t
JOIN (SELECT transaction_date, transaction_description, amount, COUNT(*)
FROM transactions
GROUP BY transaction_description, amount
HAVING count(*) > 1 ) b
ON t.transaction_description = b.transaction_description
AND t.amount = b.amount
ORDER BY t.amount ASC;

理想情况下，我希望输出类似于以下内容：

+------------------+-------------------------+--------+
| transaction_date | transaction_description | amount |
+------------------+-------------------------+--------+
| 2020-08-20       | Burger King             |  10.06 |
| 2020-08-23       | Burger King             |  10.06 |
| 2020-09-04       | Wendy's                 |   7.45 |
| 2020-09-06       | Wendy's                 |   7.45 |
+------------------+-------------------------+--------+

我走了吗？或者这甚至可能吗？提前致谢。

Answer 1

你可以使用exists:

select t.*
from mytable t
where exists (
    select 1
    from mytable t1
    where 
        t1.transaction_description = t.transaction_description
        and t1.transaction_date <> t.transaction_date 
        and t1.transaction_date >= t. transaction_date - interval 3 day
        and t1.transaction_date <= t. transaction_date + interval 3 day

如果您运行的是 MySQL 8.0，窗口日期范围内的计数是一个合理的选择：

select t.*
from (
    select t.*,
        count(*) over(
            partition by transaction_description
            order by transaction_date
            range between interval 3 day preceding and interval 3 day following 
        ) cnt
    from mytable t
) t
where cnt > 1