我正在尝试根据以前的分区提供排名

如何解决我正在尝试根据以前的分区提供排名

我正在尝试更改排名列，而不是订购 by RMSE ASC 我想要 order by RMSESum ASC（这是以前的 partition by）......当我这样做时，我收到一个错误“有序分析函数不能嵌套”对此有任何想法吗？

Select material,plant,fiscper,dmd_final_Fcst,sum_final_dmd_qty,row_number() Over (partition by material,plant order by **RMSE ASC**) as **Ranking**,Sum(RMSE) over (partition by material,plant) as **RMSESum**,Power(Fcst_er,2) As Sq_er,dmd_final_fcst - sum_final_dmd_qty As Fcst_Er,Sqrt(MSE) As RMSE,Avg(Sq_er) as MSE 
from PRD_DWH_PURCHASING.HistoricalForecastVal
group by 1,2,3,4,5;

解决方法

您应该按如下方式使用子查询：

SELECT T.*,row_number() Over (partition by material,plant order by RMSESum ASC) as Ranking
FROM
(Select material,plant,fiscper,dmd_final_Fcst,sum_final_dmd_qty,Sum(RMSE) over (partition by material,plant) as RMSESum,Power(Fcst_er,2) As Sq_er,dmd_final_fcst - sum_final_dmd_qty As Fcst_Er,Sqrt(MSE) As RMSE,Avg(Sq_er) as MSE
  from PRD_DWH_PURCHASING.HistoricalForecastVal
 group by 1,2,3,4,5) t;

,

这样做：

select * from (Select material,plant order by RMSE ASC) as **Ranking**,Avg(Sq_er) as MSE 
from PRD_DWH_PURCHASING.HistoricalForecastVal)
order by RMSESum
;

,

您可以在不使用子查询的情况下混合使用窗口函数和聚合：

select material,plant order by Sqrt(MSE)) as Ranking
       sum(RMSE) over (partition by material,Avg(Sq_er) as MSE 
from PRD_DWH_PURCHASING.HistoricalForecastVal
group by 1,5;

但我仍然认为这行不通。你有一堆未聚合的列。我猜你根本不需要聚合而且真的想要：

select material,Avg(Sq_er) over (partition by material,sum_final_dmd_qty) as MSE 
from PRD_DWH_PURCHASING.HistoricalForecastVal;