SQL（用于Teradata）

Question

如果我有如下数据集：

1/01/2014   a
2/01/2014   a
3/01/2014   a
4/01/2014   b
5/01/2014   b
6/01/2014   b
7/01/2014   b
8/01/2014   a
9/01/2014   a
10/01/2014  a
11/01/2014  a
12/01/2014  a
13/01/2014  a

我将如何获得如下所示的输出：

letter  min     max
a   1/01/2014   3/01/2014
b   4/01/2014   7/01/2014
a   8/01/2014   13/01/2014

Answer 1

Teradata支持窗口功能。 您需要计算一个组标识符。 一种方法是行号的不同：

select letter, min(date), max(date)
from (select t.*,
             (row_number() over (order by date) -
              row_number() over (partition by letter order by date)
             ) as grp
      from t
     ) t
group by letter, grp;