data_job/sql/ads/order_retention.sql
2025-03-04 10:24:36 +08:00

42 lines
2.1 KiB
SQL

-- 处理逻辑说明
-- 1.用到了first_user 和 lately_order_user 这两个表来计算ltv
-- first_user是用户首次登陆时间
-- lately_user是用户最后一次登陆时间
-- 2.用户首次登陆去关联用户最后一次登陆时间他们的时间差就是用户的流失时间
INSERT INTO ads.order_retention ( relation_id,prodid,ds,num,ltv1,ltv2,ltv3,ltv4,ltv5,ltv6,ltv7,ltv14,ltv30,ltv60,ltv90,ltv150,ltv300)
select
t1.relation_id
,t1.prodid
,t1.ds
,count(distinct t1.user_id) AS num
,SUM(CASE WHEN t2.ds_date-t1.ds_date = 1 then 1 ELSE 0 END) AS ltv1
,SUM(CASE WHEN t2.ds_date-t1.ds_date = 2 then 1 ELSE 0 END) AS ltv2
,SUM(CASE WHEN t2.ds_date-t1.ds_date = 3 then 1 ELSE 0 END) AS ltv3
,SUM(CASE WHEN t2.ds_date-t1.ds_date = 4 then 1 ELSE 0 END) AS ltv4
,SUM(CASE WHEN t2.ds_date-t1.ds_date = 5 then 1 ELSE 0 END) AS ltv5
,SUM(CASE WHEN t2.ds_date-t1.ds_date = 6 then 1 ELSE 0 END) AS ltv6
,SUM(CASE WHEN t2.ds_date-t1.ds_date = 7 then 1 ELSE 0 END) AS ltv7
,SUM(CASE WHEN t2.ds_date-t1.ds_date = 14 then 1 ELSE 0 END) AS ltv14
,SUM(CASE WHEN t2.ds_date-t1.ds_date = 30 then 1 ELSE 0 END) AS ltv30
,SUM(CASE WHEN t2.ds_date-t1.ds_date = 60 then 1 ELSE 0 END) AS ltv60
,SUM(CASE WHEN t2.ds_date-t1.ds_date = 90 then 1 ELSE 0 END) AS ltv90
,SUM(CASE WHEN t2.ds_date-t1.ds_date = 150 then 1 ELSE 0 END) AS ltv150
,SUM(CASE WHEN t2.ds_date-t1.ds_date = 300 then 1 ELSE 0 END) AS ltv300
from
"dwd"."order_user_df" t1
left join
dwd.active_df t2
on t1.relation_id=t2.relation_id
and t1.user_id=t2.user_id
AND t2.ds_date-t1.ds_date IN (1,2,3,4,5,6,7,14,30,60,90,150,300)
-- AND t2.ds_date >= t1.ds_date - 300
where t1.ds>='${300_days_later}'
group by t1.relation_id
,t1.prodid
,t1.ds
ON CONFLICT (relation_id,prodid,ds)
DO UPDATE SET num = EXCLUDED.num,ltv1 = EXCLUDED.ltv1,ltv2 = EXCLUDED.ltv2,ltv3 = EXCLUDED.ltv3,ltv4 = EXCLUDED.ltv4,ltv5 = EXCLUDED.ltv5,ltv6 = EXCLUDED.ltv6,ltv7 = EXCLUDED.ltv7,ltv14 = EXCLUDED.ltv14,ltv30 = EXCLUDED.ltv30,ltv60 = EXCLUDED.ltv60
,ltv90 = EXCLUDED.ltv90,ltv150 = EXCLUDED.ltv150,ltv300 = EXCLUDED.ltv300
;