Question

我的数据有两个字段。一个是会员ID，另一个是会员多年来的通话日期。我有兴趣找到每个会员之间的平均通话天数。我的日期只有一个字段（列）。

MembershipId    CallDates
123             01/01/2014
123             01/06/2014
123             01/15/2014
234             01/01/2014
234             01/15/2014
345             01/06/2014
456             01/06/2014

我的查询：

WITH OrderDates AS 
  ( 
    SELECT 
        SC_CALL_MBR_ID, 
        ROW_NUMBER() OVER (partition by SC_CALL_MBR_ID ORDER BY SC_DT DESC) AS RowNumber, 
        SC_DT 
    FROM [D3Reporting].[dbo].[ASC_ARCH_CALL] 
    WHERE sc_dt >'2015-06-17 00:00:00.0000000'
  ) 

SELECT 
    tab1.SC_CALL_MBR_ID, 
    avg(DATEDIFF(DD, tab2.SC_DT ,tab1.SC_DT)) AS 'AverageDate' 
FROM 
    OrderDates tab1 
     LEFT JOIN 
    OrderDates tab2 ON 
        tab2.RowNumber = tab1.RowNumber + 1 
GROUP BY tab1.SC_CALL_MBR_ID

Answer 1

为每个成员获取不同的通话日期，然后从平均结果中取出平均值。

Select   t.MemberId
        ,Avg(CallDate) As CallPerMember
From    (
            Select   Distinct
                     MemberId
                    ,CallDate
            From    Membership As m With (Nolock)
        ) As t
Group By t.MemberId

Answer 2

执行此操作的一种方法是通过自联接获取先前日期，计算通话日期之间的差异，然后计算平均值。请参阅下面的第一个查询。

该问题被标记为SQL Server 2008，因此您无法使用LAG和LEAD。我在下面添加了另一个示例，它将LAG用于使用更新版本的用户。

declare @Membership table
(
    id int,
    calldate datetime
)

-- generate some test data    
insert into @Membership
select 1, '1/1/2015'
union all select 1, '1/2/2015'
union all select 1, '1/5/2015'
union all select 1, '2/1/2015 8:00 AM' -- test using a time value
union all select 1, '2/1/2015 9:00 AM'
union all select 2, '1/1/2015'
union all select 2, '1/14/2015'
union all select 2, '2/14/2015'


-- This method should work on older SQL Server versions
-- Number the membership/call date rows.
;with CallTimesOrdered (num, id, calldate) as
(
select
    ROW_NUMBER() over (partition by id order by calldate),
    id,
    calldate
    from @Membership
),
-- Join the numbered rows back to each other.
AvgCallTimes (id, timespan) as
(
    select
        CurrentDate.id,
        datediff(d, PriorDate.calldate, CurrentDate.calldate)
    from CallTimesOrdered CurrentDate
    inner join CallTimesOrdered PriorDate
        on PriorDate.num = CurrentDate.num - 1
        and PriorDate.id = CurrentDate.id
)
-- Compute the average days between calls by membership ID.
select id, avg(timespan) as AvgTimeBetweenCalls
from AvgCallTimes
group by id

-- This version works with SQL 2012 and higher versions.
-- It uses LAG function.
-- First, calculate the difference in days between the current and the prior dates.
;with AvgCallTimes (id, avgtimespan) as
(
select
    id,
    datediff(d, lag(calldate, 1, null) over (partition by id order by calldate), calldate)
from @Membership
)
-- Calculate the average time span in days by membership ID.
select id, avg(avgtimespan) as AvgTimeBetweenCalls
from AvgCallTimes
group by id

在使用下面的测试数据的两种情况下，我得到以下结果：

ID  AvgTimeBetweenCalls
1   7
2   22

如何在SQL中的字段中查找日期（日期时间）之间的平均值（以天为单位）？

2 个答案: