将当月销售额与上月销售额进行比较

Question

获取测试数据的代码：

create table SalesCalls
(
    EmpId INT NOT NULL,
    EmpName nvarchar(20),
    month INT,
    Year INT,
    CallsMade INT
)
GO

Insert into SalesCalls values
(1,'ABC',12,2018,10),
(1,'ABC',1,2019,15),
(1,'ABC',2,2019,20),
(2,'DEF',12,2018,12),
(2,'DEF',1,2019,14),
(2,'DEF',2,2019,26)
GO

objective是比较某员工当月销售额与该员工上月销售额的百分比变化。使用以下查询实现：

With SalesCTE as 
(
    select EmpId,EmpName,
        Month As CurrentMonth,
        Year as CurrentMonthYear,
        Case When month = 1 then 12 Else (Month-1) End AS PrevMonth,
        Case when month = 1 then (Year - 1) Else Year End As PrevMonthYear,
        CallsMade 
    from SalesCalls
)
select 
    S1.EmpId, S1.EmpName, S1.CurrentMonth, S1.CurrentMonthYear, S1.CallsMade as CurrentMonthCalls,
    S2.CurrentMonth as PrevMont,
    S2.CurrentMonthYear as PrevMonthYear,
    S2.CallsMade as PrevMonthCalls,
    ( CONVERT(numeric(5,2),S1.CallsMade) / S2.CallsMade) * 100 As PercentageChange
from SalesCTE S1
JOIN SalesCTE S2 ON S1.EmpId = S2.EmpId
    AND S1.PrevMonth = S2.CurrentMonth   
    AND S1.PrevMonthYear = S2.CurrentMonthYear
ORDER BY S1.EmpId, S1.CurrentMonth, S1.CurrentMonthYear

以上查询一直有效，直到同一个月的员工没有冗余记录。

但后来来自多个来源的数据进来了，一个员工 table 可以在同一个月有多个记录并且它仍然有效。因为员工可能会以不同的方式拨打电话。例如，下面的记录被插入到 table:

Insert into SalesCalls values
(1,'ABC',1,2019,1)

上面的查询在比较当前月份的 SalesCalls 与上个月时运行良好，但现在不再有效。

用例的第 2 阶段：因此，为了解决这个问题，我构建了一个包含聚合数据的中间温度 table。使用的查询是：

Select EmpId, EmpName, month, Year, SUM(CallsMade) as CallsMade 
into #SalesCalls
from SalesCalls
group by EmpId, EmpName, month, Year

现在 CTE 中的 SalesCalls table 被替换为 #SalesCalls 然后上面的查询工作正常。

但是这个#SalesCallstable每次都需要drop重新创建才能看到最新的对比数据

问题是，是否可以仅使用单个查询而不使用中间临时 tables 或视图来获取比较数据。

Answer 1

只需使用window函数：

select EmpId, EmpName, month, Year,
       sum(CallsMade) as CallsMade,
       (case when lag(year * 12 + month) over (partition by empId order by year, month) = year * 12 + month - 1
             then lag(sum(callsMade)) over (partition by empId order by year, month)
        end) as prevMonthCalls,
       (case when lag(year * 12 + month) over (partition by empId order by year, month) = year * 12 + month - 1
             then callsMade * 100.0 / lag(sum(callsMade) over (partition by empId order by year, month)
        end) as as perentageChange
from SalesCalls
group by EmpId, EmpName, month, Year;

根本不需要 joins、CTE、子查询或临时表。

Answer 2

最简单的解决方案之一：

select EmpId, EmpName, month, Year,
       sum(CallsMade) as CallsMade,
        lag(sum(callsMade)) over (partition by empId order by year, month) AS prevMonthCalls,
        sum(CallsMade) * 100.0 / lag(sum(CallsMade)) over (partition by empId order by year, month) as PercentageChange
from SalesCalls
group by EmpId, EmpName, month, Year
order by EmpId,  Year,month;

将当月销售额与上月销售额进行比较

Compare current month sales with previous month sales

sql

sql-server

tsql

sql-server-2019