原始表（order_tbl），按dt分區，每天分區保存全量訂單
(20200321分區)
order_id | address | trade_time | dt |
1 | 天津 | 20200320 | 20200321 |
2 | 北京 | 20200320 | 20200321 |
3 | 北京 | 20200319 | 20200321 |
4 | 天津 | 20200319 | 20200321 |
5 | 北京 | 20200321 | 20200321 |
6 | 20200321 | 20200321 |
(20200320分區)
order_id | address | trade_time | dt |
1 | 20200320 | 20200320 | |
2 | 北京 | 20200320 | 20200320 |
3 | 北京 | 20200319 | 20200320 |
4 | 20200319 | 20200320 |
使用SQL實現，每日訂單量，當日未填地址訂單數，1日後未填地址數量，2日後未填地址數量，3日後未填地址數量
解釋: 【1日後未填地址數量】，指某日的訂單中，下一天未填地址訂單的量
例如：trade_time=20200320，未填地址訂單1個，【1日後未填地址數量】要統計分區dt=20200321下trade_time=20200320在的未填地址訂單0個
結果表
trade_time | order_cnt | no_address_cnt | 1_day_later | 2_day_later | 3_day_later | 4_day_later | 5_day_later |
20200323 | 300 | 50 | 0 | 0 | 0 | 0 | 0 |
20200322 | 400 | 80 | 50 | 0 | 0 | 0 | 0 |
20200321 | 333 | 55 | 41 | 32 | 0 | 0 | 0 |
…… | |||||||
20200312 |
with order_tbl as ( -- 20200320分區 select 1 as order_id, '天津' as address, 20200319 as trade_time, 20200320 as dt union all select 2 as order_id, null as address, 20200319 as trade_time, 20200320 as dt union all select 3 as order_id, '北京' as address, 20200320 as trade_time, 20200320 as dt union all select 4 as order_id, '北京' as address, 20200320 as trade_time, 20200320 as dt union all select 5 as order_id, null as address, 20200320 as trade_time, 20200320 as dt union all select 6 as order_id, null as address, 20200320 as trade_time, 20200320 as dt union all select 7 as order_id, null as address, 20200320 as trade_time, 20200320 as dt union all -- 20200321分區 select 1 as order_id, '天津' as address, 20200319 as trade_time, 20200321 as dt union all select 2 as order_id, '天津' as address, 20200319 as trade_time, 20200321 as dt union all select 3 as order_id, '北京' as address, 20200320 as trade_time, 20200321 as dt union all select 4 as order_id, '北京' as address, 20200320 as trade_time, 20200321 as dt union all select 5 as order_id, '北京' as address, 20200320 as trade_time, 20200321 as dt union all select 6 as order_id, null as address, 20200320 as trade_time, 20200321 as dt union all select 7 as order_id, null as address, 20200320 as trade_time, 20200321 as dt union all select 8 as order_id, '北京' as address, 20200321 as trade_time, 20200321 as dt union all select 9 as order_id, null as address, 20200321 as trade_time, 20200321 as dt union all select 10 as order_id, null as address, 20200321 as trade_time, 20200321 as dt union all -- 20200322分區 select 1 as order_id, '天津' as address, 20200319 as trade_time, 20200322 as dt union all select 2 as order_id, '天津' as address, 20200319 as trade_time, 20200322 as dt union all select 3 as order_id, '北京' as address, 20200320 as trade_time, 20200322 as dt union all select 4 as order_id, '北京' as address, 20200320 as trade_time, 20200322 as dt union all select 5 as order_id, '北京' as address, 20200320 as trade_time, 20200322 as dt union all select 6 as order_id, '北京' as address, 20200320 as trade_time, 20200322 as dt union all select 7 as order_id, null as address, 20200320 as trade_time, 20200322 as dt union all select 8 as order_id, '北京' as address, 20200321 as trade_time, 20200322 as dt union all select 9 as order_id, null as address, 20200321 as trade_time, 20200322 as dt union all select 10 as order_id, null as address, 20200321 as trade_time, 20200322 as dt union all select 11 as order_id, '北京' as address, 20200322 as trade_time, 20200322 as dt union all select 12 as order_id, null as address, 20200322 as trade_time, 20200322 as dt union all select 13 as order_id, null as address, 20200322 as trade_time, 20200322 as dt ) select trade_time, order_cnt, 1_day_later, 2_day_later, 3_day_later from ( select trade_time, order_cnt, no_address_cnt, LEAD(no_address_cnt, 1, 0 ) over (partition by trade_time order by dt asc) as 1_day_later, LEAD(no_address_cnt, 2, 0 ) over (partition by trade_time order by dt asc) as 2_day_later, LEAD(no_address_cnt, 3, 0 ) over (partition by trade_time order by dt asc) as 3_day_later, row_number () over(partition by trade_time ORDER BY dt) as rn from ( select dt, trade_time, count(1) as order_cnt, sum(case when address is null then 1 else 0 end) as no_address_cnt from order_tbl group by dt, trade_time ) a ) b where rn = 1 order by trade_time desc