SQL Server 数据库删除重复的趋势值，但保留第一个和最后一个

Question

我有一个充满温度值的大型数据库。 问题是系统保存了大量重复值，现在SQL Server 数据库已满（Express 版）。

部分数据样本，使用查询

SELECT [PointID], [Time], [Data] 
FROM TrendSamples 
WHERE pointid = 13 
ORDER BY time DESC

结果：

PointID Time       Data
---------------------------------------------
13  2020-01-02 07:29:01.077 22,2999992370605
13  2020-01-02 07:28:50.937 22,5
13  2020-01-02 07:28:05.230 22,2999992370605
13  2020-01-02 07:27:55.090 22,3999996185303 
13  2020-01-02 07:27:04.510 22,3999996185303
13  2020-01-02 07:26:13.443 22,3999996185303
13  2020-01-02 07:25:22.580 22,3999996185303
13  2020-01-02 07:24:31.340 22,3999996185303
13  2020-01-02 07:23:40.370 22,3999996185303
13  2020-01-02 07:22:49.460 22,3999996185303
13  2020-01-02 07:21:59.160 22,3999996185303
13  2020-01-02 07:21:08.483 22,3999996185303
13  2020-01-02 07:20:17.713 22,3999996185303
13  2020-01-02 07:19:26.710 22,3999996185303
13  2020-01-02 07:18:35.283 22,3999996185303
13  2020-01-02 07:17:44.250 22,3999996185303
13  2020-01-02 07:16:53.463 22,3999996185303
13  2020-01-02 07:16:02.367 22,3999996185303
13  2020-01-02 07:15:11.083 22,3999996185303
13  2020-01-02 07:14:19.987 22,3999996185303
13  2020-01-02 07:13:29.230 22,3999996185303
13  2020-01-02 07:12:38.197 22,3999996185303
13  2020-01-02 07:11:47.957 22,3999996185303
13  2020-01-02 07:10:57.033 22,3999996185303
13  2020-01-02 07:10:06.293 22,3999996185303
13  2020-01-02 07:09:15.183 22,3999996185303
13  2020-01-02 07:08:24.083 22,3999996185303
13  2020-01-02 07:07:33.237 22,3999996185303
13  2020-01-02 07:06:42.140 22,3999996185303
13  2020-01-02 07:05:51.557 22,3999996185303
13  2020-01-02 07:05:00.787 22,3999996185303
13  2020-01-02 07:04:09.707 22,3999996185303
13  2020-01-02 07:03:18.970 22,3999996185303
13  2020-01-02 07:02:28.043 22,3999996185303
13  2020-01-02 07:01:36.930 22,3999996185303
13  2020-01-02 07:00:46.317 22,3999996185303
13  2020-01-02 06:59:55.390 22,3999996185303
13  2020-01-02 06:59:04.403 22,3999996185303
13  2020-01-02 06:58:13.103 22,3999996185303
13  2020-01-02 06:58:01.247 22,5

如您所见，有很多重复数据，其中的数据在不同时间之间根本没有变化。 有什么方法可以删除所有重复数据，但在值更改之前和之后保留第一行和最后一行。

结果我想要的是这个

PointID Time    Data
---------------------------------------------
13  2020-01-02 07:29:01.077 22,2999992370605
13  2020-01-02 07:28:50.937 22,5
13  2020-01-02 07:28:05.230 22,2999992370605
13  2020-01-02 07:27:55.090 22,3999996185303 
13  2020-01-02 06:58:13.103 22,3999996185303
13  2020-01-02 06:58:01.247 22,5

Answer 1

您可以为此使用可更新的 CTE：

with cte as (
    select 
        t.*, 
        lag(data) over(partition by pointID order by time) lag_data,
        lead(data) over(partition by pointID order by time) lead_data
    from mytable t
)
delete from cte where (data = lag_data and data = lead_data)

CTE 使用lag()和lead()来引入具有相同pointID的前一行和下一行的data值，按time排序。 然后外部查询删除data与上一条和下一条记录相同的记录。

Answer 2

假设您的表具有自动生成的带有唯一编号的 ID 列。

在 cte 表达式中使用 RowNumber() 两次，我们可以使用自动生成的 Id 列按 asc 和 desc 顺序对数据进行分区，并删除行。

with result as (
        select *, 
        Row_Number() over(partition by Data order by ID) rownumber1,
        Row_Number() over(partition by Data  order by ID desc) rownumber2
        from TrendSamples 
    )
    delete from result where rownumber1 > 1 and rownumber2 > 1