Question

I have a table called TableA.

Input:

ColumnA      ColumnB    ColumnC
jim          1788           5F
jim          2000           9F
jim          500            9F
ben          190            4H
geri          40        manny
geri          40        kelly
geri          20        joker
geri          20        jam

Output:

ColumnA     ColumnB     ColumnC
jim            2000         9F
jim            2000         NULL
ben             190         4H
geri            40          manny
geri            40          kelly
geri            40          NULL

Can someone help me with the SQL query?

The logic is the following:

I want to group the rows by ColumnA, e.g. the “gerri” group will have 4 rows. For each group, the maximum element from ColumnB is calculated , for gerri group it’s 40.
Each element in the group is analysed for ColumnB: if the element.ColumnB = maximum, the row is put in the output (if it doesn’t exist already). Otherwise, if the element.ColumnB different than the maximum, the current row is put in the output with NULL on ColumnC and MAXIMUM on ColumnB (again, if it doesn’t exist in the output already).

It’s clearer from the example I gave.

Thank you for any suggestions in advance!

Answer 1

; with maxes as (
  select ColumnA,
         ColumnB,
         ColumnC,
         max(ColumnB) over (partition by ColumnA) mx
    from tablea
)
select distinct
       ColumnA, 
       mx ColumnB, 
       case when mx = ColumnB 
            then ColumnC 
            else null 
        end ColumnC
from maxes

Example on Sql Fiddle.