Question

我正在编写一个查询，该查询计算每年的某天活跃18个月的人数。

我遇到的问题是，各年之间存在时间重叠，这导致后几年被低估，因为它们被归类为上一年。

例如，“ 2017-03-06”可以被视为2018年和2017年的活动。这是我的查询：

select case when deposit_dt between '2017-02-07' and date then '2018'
    when deposit_dt between '2016-02-07' and '2017-08-07'  then '2017'
    when deposit_dt between '2015-02-07' and '2016-08-07' then '2016'
    when deposit_dt between '2014-02-07' and '2015-08-07' then '2015'
    when deposit_dt between '2013-02-07' and '2014-08-07' then '2014'
    end as yr, count(unique(op_id))
    from activity_table
    where deposit_dt between '2013-02-07' and date
    group by deposit_dt

关于如何解决此问题的任何建议（除了每年运行一次新查询）？

Answer 1

我的感觉是，您确实需要遵循以下几点：

select
    count(case when deposit_dt between '2017-02-07' and date then 1 end) as 2018,
    count(case when deposit_dt between '2016-02-07' and '2017-08-07' then 1 end) as 2017,
    count(case when deposit_dt between '2015-02-07' and '2016-08-07' then 1 end) as 2016,
    count(case when deposit_dt between '2014-02-07' and '2015-08-07' then 1 end) as 2015,
    count(case when deposit_dt between '2013-02-07' and '2014-08-07' then 1 end) as 2014
from activity_table
where deposit_dt between '2013-02-07' and date;

请注意，按deposit_dt分组是没有意义的，因为这是用于汇总的列。

这假设您没有足够的逻辑来考虑可能重叠的日期。如果您可以提供逻辑方法来解决与多个范围相匹配的日期，则可以更新上述查询。

Answer 2

我认为这可能是您想要的。

select count(distinct [2018Users]) as [2018], count(distinct [2017Users]) as [2017],count(distinct [2016Users]) as [2016] 
from(
    select case when deposit_dt  between '2017-02-07' and date then op_id  end as [2018Users],
           case when deposit_dt  between '2016-02-07' and '2017-08-07' then op_id  end as [2017Users],
           case when deposit_dt  between '2015-02-07' and '2016-08-07' then op_id  end as [2016Users]
    from activity_table
    ) c

SQL查询重叠时间不足计数

2 个答案: