get best sales rep weekly SQL - sql

Question

I need a bit of help with a SQL Server issue.
I have 2 tables:
complete_sales_raw
(
Id int Identity(1,1) PK,
RepId int FK in sale_reps,
Revenue decimal(15,2),
Sale_date datetime2(7)
)
and
sale_reps
(
Id int Identity(1,1) PK,
RepName nvarchar(50)
)
What I need to do is get best sales rep based on the total revenue for each week, starting with 2014-06-01 and ending at current date.
Each week has 7 days and the first day is 2014-06-01.
So far I got to here:
SELECT TOP(1)
sr.RepName as RepName,
SUM(csr.Revenue) as Revenue
INTO #tmp1
FROM complete_sales_raw csr
JOIN sale_reps sr on csr.RepId = sr.Id
WHERE DATEDIFF( d,'2014-06-01', Sale_date ) BETWEEN 0 and 6
GROUP BY sr.RepName
ORDER BY 2 desc
But this only returns the best sale rep for the first week and I need it for each week.
All help is appreciated.

Answer 1

ok so, I created a week table like so
IF ( OBJECT_ID('dbo.tmp4') IS NOT NULL )
DROP TABLE dbo.tmp4
GO
Create Table tmp4(
StartDate datetime,Enddate datetime,WeekNo varchar(20)
)
DECLARE
#start_date DATETIME,
#end_date DATETIME,
#start_date1 DATETIME,
#end_date1 DATETIME
DECLARE #Table table(StartDate datetime,Enddate datetime,WeekNo varchar(20))
Declare #WeekDt as varchar(10)
SET #start_date = '2014-06-01'
SET #end_date = '2015-01-03'
Set #WeekDt = DATEPART(WEEK,#start_date)
SET #start_date1 = #start_date
While #start_date<=#end_date
Begin
--Select #start_date,#start_date+1
IF #WeekDt<>DATEPART(WEEK,#start_date)
BEGIN
Set #WeekDt = DATEPART(WEEK,#start_date)
SET #end_date1=#start_date-1
INSERT INTO tmp4 Values(#start_date1,#end_date1,DATEPART(WEEK,#start_date1))
SET #start_date1 = #start_date
END
set #start_date = #start_date+1
END
GO
and then I used Gordon's answer and made this:
SELECT t.StartDate as StartDate, sr.RepName as RepName, SUM(csr.Revenue) as Revenue,
RANK() OVER (PARTITION BY (t.StartDate) ORDER BY SUM(csr.Revenue) desc) as seqnum into tmp1
FROM tmp4 t,
complete_sales_raw csr
JOIN sale_reps sr on csr.RepId = sr.Id
WHERE DATEDIFF( d,t.StartDate, MAS_PostDate ) BETWEEN 0 and 6
GROUP BY sr.RepName, t.StartDate
SELECT * FROM tmp1
WHERE seqnum = 1
ORDER BY StartDate
which returns the best sales_rep for each week

Answer 2

You can do an aggregation to get the total sales by week. This requires some manipulation of the dates to calculate the number of weeks -- basically dividing the days by 7.
Then, use rank() (or row_number() if you only want one when there are ties) to get the top value:
SELECT s.*
FROM (SELECT tsr.RepName as RepName,
(DATEDIFF(day, '2014-06-01', MAS_PostDate ) - 1) / 7 as weeknum,
SUM(csr.Revenue) as Revenue,
RANK() OVER (PARTITION BY (DATEDIFF(day, '2014-06-01', MAS_PostDate ) - 1) / 7 ORDER BY SUM(csr.Revenue)) as seqnum
FROM complete_sales_raw csr JOIN
sale_reps sr
on csr.RepId = sr.Id
WHERE DATEDIFF(day, '2014-06-01', MAS_PostDate ) BETWEEN 0 and 6
GROUP BY sr.RepName, (DATEDIFF(day, '2014-06-01', MAS_PostDate ) - 1) / 7
) s
WHERE seqnum = 1;