在 SQL 中计算 4 天滚动平均值时出错答案

【问题标题】：Error while calculating 4 Day Rolling Average in SQL在 SQL 中计算 4 天滚动平均值时出错
【发布时间】：2018-10-03 17:12:23
【问题描述】：

我正在尝试在 SQL 中计算 4 天滚动平均值。我有以下格式的日期。我的桌子上有很多公司。我必须为每家公司分别计算滚动平均值。但是我下面的查询是在不区分公司的情况下执行滚动平均。

请更正我的查询。预期输出如下：

  select 
[date],[PX_VOLUME],[Company],
avg([PX_VOLUME]) OVER (ORDER BY [date],[Company] ROWS BETWEEN 3 PRECEDING AND CURRENT ROW) as monthrolling
into dbo.[Equity_Indicators_window]
from dbo.[Equity_Indicators_V1]
group by [Company],[date],[PX_VOLUME]
order by [date]

输入数据：

date    PX_Volume   Company
2018-06-22  7455031       ABC HOLDINGS
2018-06-22  801629        LEMAITRE VASLA
2018-06-22  23951629         CLUB CORP
2018-06-23  7455041       ABC HOLDINGS
2018-06-23  801631        LEMAITRE VASLA
2018-06-23  23951643         CLUB CORP
2018-06-24  745506        ABC HOLDINGS
2018-06-24  801666        LEMAITRE VASLA
2018-06-24  23951698         CLUB CORP
2018-06-25  7455031       ABC HOLDINGS
2018-06-25  801629        LEMAITRE VASLA
2018-06-25  23951629         CLUB CORP
2018-06-26  7455031       ABC HOLDINGS
2018-06-26  801629        LEMAITRE VASLA
2018-06-26  23951629         CLUB CORP

预期输出：

date          PX_Volume          Company    monthrolling
2018-06-22  7455031       ABC HOLDINGS    7455031
2018-06-22  801629        LEMAITRE VASLA    801629
2018-06-22  23951629         CLUB CORP      23951629
2018-06-23  7455041       ABC HOLDINGS     7455036
2018-06-23  801631        LEMAITRE VASLA    801630
2018-06-23  23951643         CLUB CORP     23951636
2018-06-24  745506        ABC HOLDINGS      5218526
2018-06-24  801666        LEMAITRE VASLA    801642
2018-06-24  23951698         CLUB CORP     23951656.67
2018-06-25  7455031       ABC HOLDINGS     5777652.25
2018-06-25  801629        LEMAITRE VASLA    801638.75
2018-06-25  23951629         CLUB CORP      23951649.75
2018-06-26  7455031       ABC HOLDINGS     5777652.25
2018-06-26  801629        LEMAITRE VASLA    801638.75
2018-06-26  23951629         CLUB CORP     23951649.75

【问题讨论】：

标签： sql sql-server moving-average

【解决方案1】：

您的问题是group by 中的px_volume。我想你打算：

select [date], [Company], sum(px_volume) as px_volume,
       avg(sum(px_volume)) over (partition by company
                                 order by [date]
                                 rows between 3 preceding and current_row
                                ) as monthrolling
into dbo.[Equity_Indicators_window]
from dbo.[Equity_Indicators_V1]
group by [Company], [date]
order by [date] ;

注意partition by company。

如果您在每个日期每个公司有一行，则不需要聚合。那将是：

select [date], Company, px_volume,
       avg(px_volume) over (partition by company
                            order by [date]
                            rows between 3 preceding and current_row
                           ) as monthrolling
into dbo.[Equity_Indicators_window]
from dbo.[Equity_Indicators_V1]
order by [date]

【讨论】：

@ArvinthKumar 。 . .令我惊讶的是，您接受了在此之后近 20 分钟发布的等效答案。
对不起，我没有检查订单。

【解决方案2】：

您的查询难以阅读在这种情况下，我建议您使用 CTE AND 在您的 over 子句中使用 partition by

with c as(
Select 
[date]
,[Company]
,AVG([PX_VOLUME]) AS PX_VOLUME
from dbo.[Equity_Indicators_V1]
group by [Company],[date]

)
SELECT [date]
      ,[PX_VOLUME]
      ,[Company]
      ,avg([PX_VOLUME]) OVER (Partition by [date],[Company] ORDER BY [date],[Company] ROWS BETWEEN 3 PRECEDING AND CURRENT ROW ) as monthrolling
into dbo.[Equity_Indicators_window]    
FROM C

尝试按步骤进行

【讨论】：

未按预期工作。感谢您抽出宝贵时间！

【解决方案3】：

试试下面的（需要在partition by子句中取公司）：

SELECT [date], 
       [PX_VOLUME], 
       [Company], 
       ROUND(AVG(CONVERT(NUMERIC(18,2), [PX_VOLUME])) OVER(PARTITION BY COMPANY ORDER BY [date] ROWS BETWEEN 3 PRECEDING AND CURRENT ROW), 2) AS monthrolling
INTO dbo.[Equity_Indicators_window]
FROM dbo.[Equity_Indicators_V1]
GROUP BY [Company], 
         [date], 
         [PX_VOLUME]
ORDER BY [date];

【讨论】：