sql Server: Rank by sum of points and order by ranking - sql-server

Question

I have a game table with these fields:
ID Name Email Points
----------------------------------
1 John john#aaa.com 120
2 Test bob#aaa.com 100
3 John john#bbb.com 80
4 Bob bob#aaa.com 50
5 John john#aaa.com 80
I want to group them by email (email Identifies that both players are the same no matter that row 2 and 4 have different names) and have also sum of points and the last entered name in the results and rank them with the heighest sum of points to the lowest
the Result I want from the sample table is:
Ranking Name Points Games_Played Average_Points
------------------------------------------------------------------------------------------
1 John 200 2 100
2 Bob 150 2 75
3 John 80 1 80
I could achieve getting ranking, sum of points, and average points but getting the last entered name I think need joining with the same table again and it seems a little wrong.
Any ideas how to do this?

Answer 1

Displaying the Name and grouping be email will cause to use e.g. MIN(Name) and lead to duplicate names.
Select Rank() over (order by Points desc) as Rank
,Name,Points,Games_Played,Average_Points
from
(
Select Min(Name) as Name,Email,Sum(Points) as Points
,Count(*) as Games_Played,AVG(Points) as Average_Points
From #a Group by Email
) a
order by Rank
SQLFiddle
in the Fiddle are two commented lines you should uncomment to see the behavior on identical results.

Answer 2

You can use Ranking Functions from SQL-Server 2005 upwards:
WITH Points
AS (SELECT Sum_Points = Sum(points) OVER (
partition BY email),
Games_Played = Count(ID) OVER (
partition BY email),
Average_Points = AVG(Points) OVER (
partition BY email),
Rank = DENSE_RANK() OVER (
Partition BY email Order By Points DESC),
*
FROM dbo.Game)
SELECT Ranking=DENSE_RANK()OVER(ORDER BY Sum_Points DESC),
Name,
Points=Sum_Points,
Games_Played,
Average_Points
FROM Points
WHERE Rank = 1
Order By Sum_Points DESC;
DEMO
Note that the result is different since i'm showing the row with the highest point in case that the email is not unique, so "Test" instead of "Bob".

Answer 3

Below are separate solutions for SQL Server 2012+, 2005 to 2008 R2, and 2000:
2012+
CREATE TABLE #PlayerPoints
( ID INT PRIMARY KEY
, Name VARCHAR(10) NOT NULL
, Email VARCHAR(20) NOT NULL
, Points INT NOT NULL);
INSERT INTO #PlayerPoints (ID, Name, Email, Points)
VALUES
(1, 'John', 'john#aaa.com', 120)
, (2, 'Test', 'bob#aaa.com', 100)
, (3, 'John', 'john#bbb.com', 80)
, (4, 'Bob', 'bob#aaa.com', 50)
, (5, 'John', 'john#aaa.com', 80)
WITH BaseData
AS
(SELECT ID
, Email
, Points
, LastRecordName = LAST_VALUE(Name) OVER
(PARTITION BY Email
ORDER BY ID DESC
ROWS UNBOUNDED PRECEDING)
FROM #PlayerPoints)
SELECT Email
, LastRecordName = MAX(LastRecordName)
, Points = SUM(Points)
, Games_Played = COUNT(*)
, Average_Points = AVG(Points)
FROM BaseData
GROUP BY Email
ORDER BY Points DESC;
2005 to 2008 R2
CREATE TABLE #PlayerPoints
( ID INT PRIMARY KEY
, Name VARCHAR(10) NOT NULL
, Email VARCHAR(20) NOT NULL
, Points INT NOT NULL);
INSERT INTO #PlayerPoints (ID, Name, Email, Points)
VALUES
(1, 'John', 'john#aaa.com', 120)
, (2, 'Test', 'bob#aaa.com', 100)
, (3, 'John', 'john#bbb.com', 80)
, (4, 'Bob', 'bob#aaa.com', 50)
, (5, 'John', 'john#aaa.com', 80)
WITH BaseData
AS
(SELECT ID
, Email
, Name
, ReverseOrder = ROW_NUMBER() OVER
(PARTITION BY Email
ORDER BY ID DESC)
FROM #PlayerPoints)
SELECT pp.Email
, LastRecordName = MAX(bd.Name)
, Points = SUM(pp.Points)
, Games_Played = COUNT(*)
, Average_Points = AVG(pp.Points)
FROM #PlayerPoints pp
JOIN BaseData bd
ON pp.Email = bd.Email
AND bd.ReverseOrder = 1
GROUP BY pp.Email
ORDER BY Points DESC;
2000
CREATE TABLE #PlayerPoints
( ID INT PRIMARY KEY
, Name VARCHAR(10) NOT NULL
, Email VARCHAR(20) NOT NULL
, Points INT NOT NULL);
INSERT INTO #PlayerPoints (ID, Name, Email, Points)
SELECT 1, 'John', 'john#aaa.com', 120
UNION ALL
SELECT 2, 'Test', 'bob#aaa.com', 100
UNION ALL
SELECT 3, 'John', 'john#bbb.com', 80
UNION ALL
SELECT 4, 'Bob', 'bob#aaa.com', 50
UNION ALL
SELECT 5, 'John', 'john#aaa.com', 80;
SELECT pp.Email
, LastRecordName = MAX(sppmi.Name)
, Points = SUM(pp.Points)
, Games_Played = COUNT(*)
, Average_Points = AVG(pp.Points)
FROM #PlayerPoints pp
JOIN
(SELECT spp.Email
, spp.Name
FROM #PlayerPoints spp
JOIN
(SELECT Email
, MaximumID = MAX(ID)
FROM #PlayerPoints
GROUP BY Email) mi
ON spp.ID = mi.MaximumID) sppmi
ON pp.Email = sppmi.Email
GROUP BY pp.Email
ORDER BY Points DESC;

Answer 4

I think this is what you need
select ROW_NUMBER() OVER (ORDER BY sum(r1.points) Desc) as Ranking,
r1.name as Name,
sum(r1.points) as Points,
r3.gplayed as 'Games Played',
r2.points 'Average Points'
from ranks r1
join (select avg(points) as points, email from ranks group by email) r2
on r1.email = r2.email
join (select email, count(*) as gplayed from ranks group by email) r3
on r1.email = r3.email
group by
r1.email,
r1.name,
r2.points,
r3.gplayed
Here is a SQL Fiddle.

Answer 5

Only the solution from #RegisteredUser seems to handle the constraint on the name. However, it requires SQL Server 2012, so here is a more general solution:
Select dense_rank() over (order by sum(points) desc) as ranking
max(case when islastid = 1 then Name end) as Name, Email, Sum(Points) as Points,
Count(*) as Games_Played, AVG(Points) as Average_Points
From (select g.*,
row_number() over (partition by email order by id desc) as islastid
from games g
) t
Group by Email;
You don't have enough information in the question to choose between rank() and dense_rank().
Also, this version is simpler relative to other versions, because you can mix window functions and aggregation functions.