Question

I have two tables from which I’m trying to run a query to return the maximum (or top) transaction for each person. I should note that I cannot change the table structure. Rather, I can only pull data.

People

+-----------+
| id | name |
+-----------+
| 42 | Bob  |
| 65 | Ted  |
| 99 | Stu  |
+-----------+

Transactions (there is no primary key)

+---------------------------------+
| person     | amount | date      |
+---------------------------------+
| 42         | 3      | 9/14/2030 |
| 42         | 4      | 7/02/2015 |
| 42         | *NULL* | 2/04/2020 |
| 65         | 7      | 1/03/2010 |
| 65         | 7      | 5/20/2020 |  
+---------------------------------+

Ultimately, for each person I want to return the highest amount. If that doesn’t work then I’d like to look at the date and return the most recent date.

So, I’d like my query to return:

+----------------------------------------+
| person_id  | name | amount | date      |
+----------------------------------------+
| 42         | Bob  | 4      | 7/02/2015 | (<- highest amount)
| 65         | Ted  | 7      | 5/20/2020 | (<- most recent date)
| 99         | Stu  | *NULL* | *NULL*    | (<- no records in Transactions table)
+----------------------------------------+

---

SELECT People.id, name, amount, date
FROM People
LEFT JOIN (
    SELECT TOP 1 person_id
    FROM Transactions
    WHERE person_id = People.id
    ORDER BY amount DESC, date ASC
)
ON People.id = person_id

I can’t figure out what I am doing wrong, but I know it’s wrong. Any help would be much appreciated.

Answer 1

You are almost there but since there are duplicate Id in the Transaction table ,so you need to remove those by using Row_number() function
Try this :

With cte as 
 (Select People,amount,date ,row_number() over (partition by People 
                                  order by amount desc, date desc) as row_num
  from Transac )
 Select * from People as a 
 left join cte as b
 on a.ID=b.People
 and b.row_num=1

The result is in Sql Fiddle

Edit: Row_number() from MSDN

Returns the sequential number of a row within a partition of a result set, 
starting at 1 for the first row in each partition.

Partition is used to group the result set and Over by clause is used

Determine the partitioning and ordering of the rowset before the
associated window function is applied.