根据后续交易对客户交易进行分组

Question

我需要一个 SQL 查询来对客户首次交易后 30 天内进行的交易进行分组。 将客户首次交易后 30 天内进行的所有连续交易视为一个组。如果在前一组的第一笔交易之后超过 30 天进行连续交易，则会开始新的组。

数据：

trans_Id	客户 ID	trans_日期
001	1101	2020-11-02
002	1101	2020-11-14
003	1101	2020-11-18
004	1101	2021-12-04
005	1101	2021-12-05
006	1101	2021-12-08
007	1101	2021-01-17
008	1101	2021-05-01
009	1101	2021-05-04
010	1102	2021-03-02
011	1102	2021-03-08
012	1102	2021-04-01
013	1102	2021-04-02
014	1102	2021-04-12
015	1102	2021-04-29
016	1102	2021-06-10
017	1102	2021-06-12

预期结果（预期如下分组）。

客户 1101 的交易 002 和 003 被分组为 1，因为它们发生在第一笔交易 001 的 30 天内。交易 004 超出了组 1 中第一笔交易的 30 天窗口，因此它启动了一个新组 group 2. 交易 005 和 006 属于交易 004 的 30 天窗口内，因此将它们分为 2。也就是说，每组开始和结束交易的时间差应该是 30 天。

trans_Id	客户 ID	trans_日期	团体
001	1101	2020-11-02	1
002	1101	2020-11-14	1
003	1101	2020-11-18	1
004	1101	2021-12-04	2
005	1101	2021-12-05	2
006	1101	2021-12-08	2
007	1101	2021-01-17	3
008	1101	2021-05-01	4
009	1101	2021-05-04	4
010	1102	2021-03-02	1
011	1102	2021-03-08	1
012	1102	2021-04-01	2
013	1102	2021-04-02	2
014	1102	2021-04-12	2
015	1102	2021-04-29	2
016	1102	2021-06-10	3
017	1102	2021-06-12	3

预期结果（预期如下分组）。

客户 1101 的交易 002 和 003 被分组为 1，因为它们发生在第一笔交易 001 的 30 天内。交易 004 超出了组 1 中第一笔交易的 30 天窗口，因此它启动了一个新组 group 2. 交易 005 和 006 属于交易 004 的 30 天窗口内，因此将它们分为 2。也就是说，每组开始和结束交易的时间差应该是 30 天。

trans_Id	客户 ID	trans_日期	团体
001	1101	2020-11-02	1
002	1101	2020-11-14	1
003	1101	2020-11-18	1
004	1101	2021-12-04	2
005	1101	2021-12-05	2
006	1101	2021-12-08	2
007	1101	2021-01-17	3
008	1101	2021-05-01	4
009	1101	2021-05-04	4
010	1102	2021-03-02	1
011	1102	2021-03-08	1
012	1102	2021-04-01	2
013	1102	2021-04-02	2
014	1102	2021-04-12	2
015	1102	2021-04-29	2
016	1102	2021-06-10	3
017	1102	2021-06-12	3

Answer 1

Other Way(ORACLE):
with jobdata as (
    select '001' as trans_Id,   '1101' as customer_id,  to_date('2020-11-02', 'YYYY-MM-DD') as trans_date from dual union all
    select '002' as trans_Id,   '1101' as customer_id,  to_date('2020-11-14', 'YYYY-MM-DD') as trans_date from dual  union all
    select '003' as trans_Id,   '1101' as customer_id,  to_date('2020-11-18', 'YYYY-MM-DD') as trans_date from dual  union all
    select '004' as trans_Id,   '1101' as customer_id,  to_date('2021-12-04', 'YYYY-MM-DD') as trans_date from dual  union all
    select '005' as trans_Id,   '1101' as customer_id,  to_date('2021-12-05', 'YYYY-MM-DD') as trans_date from dual  union all
    select '006' as trans_Id,   '1101' as customer_id,  to_date('2021-12-08', 'YYYY-MM-DD') as trans_date from dual  union all
    select '007' as trans_Id,   '1101' as customer_id,  to_date('2021-01-17', 'YYYY-MM-DD') as trans_date from dual  union all
    select '008' as trans_Id,   '1101' as customer_id,  to_date('2021-05-01', 'YYYY-MM-DD') as trans_date from dual  union all
    select '009' as trans_Id,   '1101' as customer_id,  to_date('2021-05-04', 'YYYY-MM-DD') as trans_date from dual  union all
    select '010' as trans_Id,   '1102' as customer_id,  to_date('2021-03-02', 'YYYY-MM-DD') as trans_date from dual  union all
    select '011' as trans_Id,   '1102' as customer_id,  to_date('2021-03-08', 'YYYY-MM-DD') as trans_date from dual  union all
    select '012' as trans_Id,   '1102' as customer_id,  to_date('2021-04-01', 'YYYY-MM-DD') as trans_date from dual  union all
    select '013' as trans_Id,   '1102' as customer_id,  to_date('2021-04-02', 'YYYY-MM-DD') as trans_date from dual  union all
    select '014' as trans_Id,   '1102' as customer_id,  to_date('2021-04-12', 'YYYY-MM-DD') as trans_date from dual  union all
    select '015' as trans_Id,   '1102' as customer_id,  to_date('2021-04-29', 'YYYY-MM-DD') as trans_date from dual  union all
    select '016' as trans_Id,   '1102' as customer_id,  to_date('2021-06-10', 'YYYY-MM-DD') as trans_date from dual  union all
    select '017' as trans_Id,   '1102' as customer_id,  to_date('2021-06-12', 'YYYY-MM-DD') as trans_date from dual 
)
select a.trans_Id, a.customer_id, a.trans_date 
    , dense_rank() over (partition by a.customer_id order by gap_group) as group_id
from (
    select a.*, min(a.trans_date) over (partition by a.customer_id order by customer_id, trans_date) as min_trans_date
        , a.trans_date - min(a.trans_date) over (partition by a.customer_id order by customer_id, trans_date) as gap_day
        , trunc((a.trans_date - min(a.trans_date) over (partition by a.customer_id order by customer_id, trans_date)) / 30) + 1 as gap_group
    from jobdata a
) a       
;