Top N Records Per Group SQL in Access

Top n records per group sql in access

I had a similar problem a year ago: Top 3 per group including 0

Using the same approach, this will return the latest three dates for each LoginID - you may get more than three records if there are tied dates for the same LoginID.

SELECT  PR1.LogInID, PR1.Score, PR1.[Date Taken]
FROM    Progress AS PR1
WHERE   PR1.[Date Taken] IN (
                        SELECT TOP 3 PR2.[Date Taken]
                        FROM    Progress PR2
                        WHERE   PR2.LoginID = PR1.LoginID
                        ORDER BY PR2.[Date Taken] DESC
                        )
ORDER BY    LoginID, [Date Taken]

Access top n in group

This should work for you. If it doesn't satisfy your requirements, post back what you need.
Your original desire was to have 25, so you'd simply modify the last clause to be HAVING COUNT(*) <= 25

SELECT  a.item, 
        a.category, 
        a.inventorycount, 
        COUNT(*) AS ranknumber
FROM inv AS a 
INNER JOIN inv AS b 
     ON (a.category = b.category) 
     AND (a.inventorycount <= b.inventorycount)
GROUP BY  a.category, 
          a.item, 
          a.inventorycount
HAVING COUNT(*) <= 2
ORDER BY a.category, COUNT(*) DESC

If you wanted to select more columns from the table, simply add them to the SELECT and `GROUP BY' clauses.

Only when you want to expand the "TOP n for each Category, foo, bar", then you would add those columns to the INNER JOIN clause as well.

--show the top 2 items for each category and year.
SELECT  a.item, 
        a.category, 
        a.year,
        a.inventorycount, 
        COUNT(*) AS ranknumber
FROM inv AS a 
INNER JOIN inv AS b 
     ON (a.category = b.category) 
     AND (a.year = b.year) 
     AND (a.inventorycount <= b.inventorycount)
GROUP BY  a.category, a.item, a.year, a.inventorycount
HAVING COUNT(*) <= 2
ORDER BY a.year, a.category, COUNT(*) DESC

MS Access Select top n query grouped by multiple fields

You can do something like this

SELECT StudentID, Year, Subject,  AVG(TestScore) AS AvgScore
  FROM
(
  SELECT StudentID, Year, Subject, TestScore
   FROM MyTable t
   WHERE TestID IN
  (
   SELECT TOP 3 TestID 
     FROM MyTable
    WHERE StudentID = t.StudentID
      AND Year = t.Year
      AND Subject = t.Subject
    ORDER BY TestScore DESC, TestID
  )
) q
 GROUP BY StudentID, Year, Subject
 ORDER BY StudentID, Year, Subject;

Sample output:

| STUDENTID | YEAR | SUBJECT | AVGSCORE |
|-----------|------|---------|----------|
|         1 | 2012 |       1 |       91 |
|         1 | 2012 |       2 |       84 |
|         2 | 2012 |       1 |       94 |
|         2 | 2012 |       3 |       95 |

Here is SQLFiddle demo.

Demo as usually is for SQL Server but expected to work in MS Access, maybe with minor syntactic tweaks

select top x by group in ms access

MsAccess can have a subquery as its select expression, which can be used to produce a limited rownumber:

SELECT *
FROM (
SELECT TblA.*, 
   (SELECT count(*) from Sheet1 TblB 
    where tblB.GROUP=tblA.GROUP and tblB.Index<=tblA.Index) as ROWNO
FROM Sheet1 TblA) Step1
where Step1.ROWNO<=Step1.A

Biggest caveat is that if two rows of the same group can have the same Index value.

Select top N rows for each group

If there were a constant number per group, you could do:

select i.*
from items as i inner join
     groups as g
     on i.group_id = g.id
where i.id in (select top 2 i2.id
               from items i2
               where i2.group_id = i.group_id
               order by i2.score desc
              );

Instead, you will need to enumerate the values and this is expensive in MS Access:

select i.*
from (select i.*,
             (select count(*)
              from items i2
              where i2.group_id = i.group_id and
                    (i2.score < i.score or
                     i2.score = i.score and i2.id <= i2.id
                    )
             ) as seqnum
      from items as i
     ) as i inner join
     groups as g
     on i.group_id = g.id
where i.seqnum <= g.top_count;

This logic implements the equivalent of row_number(), which is the right way to solve this problem (if your database supports it).

---

Related Topics