Question

我正在尝试计算每个ID的开始日期和结束日期之间的平均天数，只有一个陷阱...它必须是前一行/日期的平均值。我完全迷住了。

之前

id  date_start  date_end
1   3/3/2015    3/24/2015
1   3/27/2015   4/14/2015
1   4/15/2015   4/30/2015
1   5/1/2015    5/13/2015
1   5/17/2015   5/28/2015
1   7/9/2015    8/7/2015
2   3/7/2015    3/11/2015
3   2/17/2015   4/8/2015
3   4/13/2015   5/5/2015
4   3/12/2015   3/25/2015
4   3/25/2015   4/3/2015
4   4/3/2015    4/10/2015
4   4/10/2015   4/18/2015
5   5/13/2015   6/12/2015
5   6/12/2015   7/15/2015
5   4/19/2015   5/13/2015

之后

id  date_start  date_end    AVG_previous_cycle_days
1   3/3/2015    3/24/2015   NULL
1   3/27/2015   4/14/2015   21
1   4/15/2015   4/30/2015   19.5
1   5/1/2015    5/13/2015   18
1   5/17/2015   5/28/2015   16.5
1   7/9/2015    8/7/2015    15.4
2   3/7/2015    3/11/2015   NULL
3   2/17/2015   4/8/2015    NULL
3   4/13/2015   5/5/2015    50
4   3/12/2015   3/25/2015   NULL
4   3/25/2015   4/3/2015    13
4   4/3/2015    4/10/2015   11
4   4/10/2015   4/18/2015   9.67
5   5/13/2015   6/12/2015   NULL
5   6/12/2015   7/15/2015   30
5   4/19/2015   5/13/2015   29

Answer 1

您可以用累计总和除以累计计数来进行计算-但您必须减去当前值：

select t.*,
       ( (sum(diff) over (partition by id order by date_start) - diff) /
         (count(*) over (partition by id order by date_start) - 1)
       ) as avg_through_previous
from t cross apply
     (values (datediff(day, date_start, date_end)) v(diff);

实际上，您也可以使用windowing子句来表达这一点：

select t.*,
       avg(diff) over (partition by id
                       order by date_start
                       rows between unbounded preceding and 1 preceding
                      ) as avg_through_previous
from t cross apply
     (values (datediff(day, date_start, date_end)) v(diff);

按ID的两个日期之间的平均天数（但仅适用于当前行之前的行）

1 个答案: