对 Teradata 选择的非聚合值运行查询时出错，必须是关联组的一部分

Question

 My Query:
 SELECT Case When msa_high>=50 AND msa_high<=60 then 'low'
             When msa_high>60 AND msa_high>=70 then 'medium'
             When msa_high>70 then 'high' end as edu_lv , Sum(rev)/Sum(S_days) AS avg_sales                                        
 from store_msa left join(Select
 (EXTRACT(YEAR FROM t.saledate)||'-'||
 EXTRACT(MONTH FROM t.saledate) ||'-'||
 t.store ) AS mys,COUNT(DISTINCT t.saledate) AS S_days, sum(t.amt) AS rev
 From trnsact t
 Where (t.saledate)<'2005-08-01' AND t.stype='P' AND mys IN
(SELECT
(EXTRACT(YEAR FROM t.saledate)||'-'||
EXTRACT(MONTH FROM t.saledate) ||'-'||
t.store) AS mys
FROM trnsact t
HAVING COUNT(DISTINCT t.saledate)>20
GROUP BY mys,t.store)) AS rev
ON store_msa.store=rev.store
Group By edu_lv,avg_sales
Order By avg_sales;

有人可以指出丢失了哪些参数以找到解决此错误的方法，其次是哪个 Group By 语句产生了此错误。

Answer 1

有两个问题，看我的内联 cmets（格式化源代码使其可读后）

SELECT 
   Case 
      When msa_high>=50 
         AND msa_high<=60 
      then 'low'
      When msa_high>60 
         AND msa_high>=70 
      then 'medium'
      When msa_high>70 
      then 'high' 
   end as edu_lv , 
   Sum(rev)/Sum(S_days) AS avg_sales
from 
   store_msa 
left join 
  ( Select
      (EXTRACT(YEAR FROM t.saledate)|| '-'||
         EXTRACT(MONTH FROM t.saledate) || '-'||
         t.store ) AS mys, 
      COUNT(DISTINCT t.saledate) AS S_days, 
      sum(t.amt) AS rev
   From 
      trnsact t
   Where (t.saledate)<'2005-08-01' 
      AND t.stype='P' 
      AND mys IN
       ( SELECT
            (EXTRACT(YEAR FROM t.saledate)|| '-'||
              EXTRACT(MONTH FROM t.saledate) || '-'||
              t.store) AS mys
         FROM 
            trnsact t
         HAVING 
            COUNT(DISTINCT t.saledate)>20
         GROUP BY 
            mys, 
            t.store 
       ) 
    -- ***** missing GROUP BY, probably mys, store again? *****
  ) AS rev ON store_msa.store=rev.store
Group By 
   edu_lv, 
   avg_sales  -- ***** remove this, can't group by an aggregated column *****
Order By 
   avg_sales;