blank entries separately counted within a "partition by" clause - sql-server

Question

I am working with Sql-Server 2008.
I have a table (MyTable) which contains two columns: IDCustomer and PhoneNumber. Some of the IDCustomers do not have a PhoneNumber, i.e. the corresponding PhoneNumber entry is blank (PhoneNumber is a varchar variable).
Here I give the first entries of my Table:
IDCUstomer PhoneNumber
22
13 911
10 118
8
17 112
.... ....
I am evaluating how many times every distinct Phone Number appears using this statement:
select
PhoneNumer,
RN = ROW_NUMBER() OVER (PARTITION BY PhoneNumber ORDER BY PhoneNumber ASC)
FROM MyTable
I am intentionally not using
select PhoneNumber,
count(PhoneNumber)
from MyTable
group by PhoneNumber
because in order to achieve my final result (which is not the topic of the question) I need to use the former expression.
My question is on the result obtained using the former expression (the one with the partition by). In fact, I would expect this result:
PhoneNumber RN
2
112 1
118 1
911 1
because I know I will obtain it using the query with the group by.
But instead I get:
PhoneNumber RN
1
2
112 1
118 1
911 1
so it looks like the blank rows are separately and progressively counted. I have checked that the same happens with more than two blank entries. For istance, if I have ten blank PhoneNumbers the result of the first query is: ten blank entries in the first column and RN growing from 1 to 10 in the second column.
Hence, I would like to ask you if you know why the results are not the ones I was expecting. Am I not obtaining the expected results since I am missing something or making any mistake?
Thank you in advance.

Answer 1

Try This it may helps you
;WITH cte(IDCUstomer,PhoneNumber)
AS
(
SELECT 22, NULL UNION ALL
SELECT 13, 911 UNION ALL
SELECT 10, 118 UNION ALL
SELECT 8, NULL UNION ALL
SELECT 17, 112
)
SELECT ISNULL(CAST(PhoneNumber AS VARCHAR(10)), '') AS PhoneNumber
,RN
FROM (
SELECT PhoneNumber
,RN = ROW_NUMBER() OVER (
PARTITION BY PhoneNumber ORDER BY IDCUstomer DESC
)
,ROW_NUMBER() OVER (
ORDER BY (
SELECT NULL
)
) - 1 AS Seq
FROM cte
) DT
WHERE dt.Seq > 0
Result
PhoneNumber RN
2
112 1
118 1
911 1

Answer 2

If you need to stick with partitions instead of GROUP BY then you will need to change ROW_NUMBER to COUNT(IDCUSTOMER) OVER (PARTITION BY PhoneNumber)

Answer 3

As explained int he comments...
declare #table table (IDCUstomer int, Phonenumber varchar(8))
insert into #table
values
(1,' '),
(2,' '),
(3,'123'),
(4,'456')
select
Phonenumber,
ROW_NUMBER() over (partition by Phonenumber order by Phonenumber) as RN
from #table
group by Phonenumber
If you have NULL values in the data you can use this
declare #table table (IDCUstomer int, Phonenumber varchar(8))
insert into #table
values
(1,' '),
(2,' '),
(3,'123'),
(4,'456'),
(5,NULL)
select
Phonenumber,
ROW_NUMBER() over (partition by Phonenumber order by Phonenumber) as RN
from
(select isnull(Phonenumber,'') as Phonenumber from #table) x
group by Phonenumber