sql - 我必须合并两个不同的表 sym33 和 sym，然后检查 max svcdate 和 min svcdate 的差异，这应该大于 90 天

Question

create temporary table sand_scipher_ds_db.public.Rf as (
  select PATIENTID, max(SVCDATE)-min(SVCDATE) as days from symphony_hvh_db.sym33.dx

  union all
  
  select PATIENTID, max(SVCDATE)-min(SVCDATE) as days from symphony_hvh_db.sym.dx
  group by PATIENTID
  having days>90
  limit 100
);
select * from sand_scipher_ds_db.public.Rf limit 100

错误： SQL 编译错误：[DX.PATIENTID] 不是有效的 group by 表达式

Answer 1

您需要在您的两边都有一个 group by，UNION ALL因为这失败了，因为您使用的是聚合函数sym33.dx而不指定 group by。因此，PATIENTID 不是第一个选择中的有效 group by 表达式。

create temporary table sand_scipher_ds_db.public.Rf as (
  select PATIENTID, max(SVCDATE)-min(SVCDATE) as days from symphony_hvh_db.sym33.dx
  group by PATIENTID
  having days>90

  union all
  
  select PATIENTID, max(SVCDATE)-min(SVCDATE) as days from symphony_hvh_db.sym.dx
  group by PATIENTID
  having days>90
  limit 100
);
select * from sand_scipher_ds_db.public.Rf limit 100

您也可能想合并所有然后分组。您需要像这样进行子选择：

create temporary table sand_scipher_ds_db.public.Rf as (
  select a.PATIENTID, max(a.SVCDATE)-min(a.SVCDATE) as days from
  ( 
    select PATIENTID, SVCDATE from symphony_hvh_db.sym33.dx
    union all
    select PATIENTID, SVCDATE from symphony_hvh_db.sym.dx
  ) a
  group by PATIENTID
  having days>90
  limit 100
);
select * from sand_scipher_ds_db.public.Rf limit 100