Question

我在SQL Server 2016数据库中具有下表：

      Time                  Identifier  Category
1.    2821-01-04 17:41.0    1002        A1001
2.    2821-01-05 17:41.0    1002        A1001
3.    2821-01-06 17:42.0    1001        A1001
4.    2821-01-07 17:44.0    1002        A1001
5.    2821-01-08 17:44.0    1002        A1001
6.    2821-01-09 17:44.0    1002        A1001
7.    2821-01-10 17:44.0    1002        A1001
8.    2821-01-11 17:44.0    1002        A1001
9.    2821-01-12 17:44.0    1002        A1001
10.   2821-01-13 17:44.0    1002        A1001
11.   2821-01-14 17:45.0    1001        A1001
12.   2821-01-15 17:46.0    1002        A1001
13.   2821-01-16 17:46.0    1002        A1001
14.   2821-01-17 17:46.0    1002        A1001
15.   2821-01-18 17:46.0    1002        A1001
16.   2821-01-19 17:46.0    1002        A1001
17.   2821-01-20 17:46.0    1002        B1001
18.   2821-01-21 17:46.0    1002        B1001
19.   2821-01-22 17:46.0    1002        B1001
20.   2821-01-23 17:46.0    1001        B1001
21.   2821-01-24 17:48.0    1002        B1001
22.   2821-01-25 17:48.0    1002        B1001
23.   2821-01-26 17:48.0    1002        B1001
24.   2821-01-27 17:48.0    1002        B1001
25.   2821-01-28 17:48.0    1001        B1001
26.   2821-01-29 17:52.0    1002        B1001
27.   2821-01-30 17:52.0    1002        B1001
28.   2821-01-31 17:52.0    1002        B1001
29.   2821-02-01 17:52.0    1002        B1001

我需要的是一个SQL查询，该查询获取任意2个连续行之间的时间差，对于每个[Category]，第一个是[Identifier] = 1001，第二个是[Identifier] = 1002（也就是说，仅应针对同一类别进行计算）。要考虑的连续记录对为：记录[3,4]和[11,12]和[20,21]和[25,26]。首先将[identifier]设为1001，将连续的[identifier]设为1002，将其视为一对。因此，我们首先需要识别这些对。仅在同一[Category]中的对是相关的。属于不同类别的对应该被忽略。

我需要知道这些对之间的差异是1秒，2秒，2秒...最多5秒。

所需的输出：

Num of seconds  Category    Count
1               A1001       1
2               A1001       1
3               A1001       0
4               A1001       0
5               A1001       0
1               B1001       0
2               B1001       1
3               B1001       0
4               B1001       1
5               B1001       0

Answer 1

我相信以下内容将带您进入球场。

SELECT *
FROM
    (
        SELECT 
            [time],
            [identifier],
            [category],
            CASE WHEN identifier = 1001 THEN DATEDIFF(SECOND, [time], LEAD([time]) OVER (PARTITION BY Category ORDER BY [time], identifier)) END as TimeDiff
        FROM yourtable
    )timesub
WHERE identifier = 1001
ORDER BY category, TimeDiff

T-SQL查询（根据特定条件分析表中的连续行）

1 个答案: