如何删除除一列之外具有相同数据的重复行

Question

我正在尝试删除重复的行，但保留最近更新的行。

这是我的数据集的插图：

我正在尝试删除两行较旧的数据（在本例中为 11/19 - 以黄色突出显示）。

这是文本数据集：

employee    ID      punch_start          punch_end              punch_hours   date_load
John Doe    276567  Sep 30 2023  2:50PM  Oct  1 2023  6:00AM    15.16666667   11/19/23 2:45 PM
Jane Doe    140037  Sep 30 2023 10:00PM  Oct  1 2023  7:05AM    9.083333333   11/19/23 2:45 PM
John Doe    276567  Sep 30 2023  2:50PM  Oct  1 2023  6:00AM    15.16666667   11/20/23 2:45 PM
Jane Doe    140037  Sep 30 2023 10:00PM  Oct  1 2023  7:05AM    9.083333333   11/20/23 2:45 PM

我尝试使用这段代码（基于这个

post）：

with todelete as (

      select  
      ,[employee]
      ,[ID]
      ,[punch_start]
      ,[punch_end]
      ,[punch_hours]
      ,row_number() over

      (
       partition by 
       [employee]
      ,[ID]
      ,[punch_start]
      ,[punch_end]
      ,[punch_hours]
      order by [date_load] desc) as seqnum
      from  [dbo].[dataset]
     )
    select * from todelete where seqnum > 1;
    delete from todelete where seqnum > 1;

但是结果（选择*）是：

我们如何修改代码以区分/仅选择具有较旧 [date_load] 的行？

Answer 1

也许 SQL Server 中的 DELETE 有足够的语法来支持这一点，但我通常在这里进行 MERGE。

merge into [dbo].[dataset] as tgt using (
    select *  
      from (
              select  
                  ,[employee]
                  ,[ID]
                  ,[punch_start]
                  ,[punch_end]
                  ,[punch_hours]
                  ,[date_load]
                  ,row_number() over

                  (
                   partition by 
                   [employee]
                  ,[ID]
                  ,[punch_start]
                  ,[punch_end]
                  ,[punch_hours]
                  order by [date_load] desc) as seqnum
              from  [dbo].[dataset]
            ) ranked
     where seqnum > 1
     ) as mrg
    ON (
       tgt.employee = mrg.employee and
       tgt.id = mrg.id and
       tgt.punch_start = mrg.punch_start and
       tgt.punch_end = mrg.punch_end and
       tgt.punch_hours = mrg.punch_hours and 
       tgt.date_load = mrg.date_load
       )
    WHEN MATCHED THEN DELETE;