Finding minimum from column value between group of rows - sql

Question

I have a table with six columns:
EKey ABC XYZ DOB My_Min Row_Num
---- ---- --- ---- ------ -------
101 AB10 123 1946 100 1
103 AB10 123 1946 200 2
201 TN10 456 1955 150 1
220 TN10 456 1955 240 2
216 TN10 456 1955 80 3
214 TN10 456 1955 80 4
I want to compute a new column Required_Min which should have the values as shown below:
EKey ABC XYZ DOB My_Min Row_Num Required_Min
---- ---- --- ---- ------ ------- ------------
101 AB10 123 1946 100 1 100
103 AB10 123 1946 200 2 100
201 TN10 456 1955 150 1 80
220 TN10 456 1955 240 2 80
216 TN10 456 1955 80 3 80
214 TN10 456 1955 80 4 80
Im using SQL for this, i.e., SSMS. Please help.

Answer 1

You can JOIN the table back to itself using the MIN aggregate:
select t.ekey, t.abc, t.xyz, t.dob, t.my_min, t.rownum,
t2.required_min
from yourtable t
join (
select ekey, min(my_min) required_min
from yourtable
group by ekey
) t2 on t.ekey = t2.ekey

Answer 2

If your uniqueness is based on ABC+XYZ+DOB, you can do the following:
with TableMin as
(select ABC,XYZ,DOB,min(My_min) as My_min
group by ABC,XYZ,DOB
from Table1)
update t1 set t1.Required_Min=tm.My_min
from Table1 t1 inner join TableMin tm on t1.ABC=tm.ABC and t1.XYZ=tm.XYZ and t1.DOB=tm.DOB
But this join is quite hard for server with big amount of data. If ABC+XYZ in unique key (or even ABC), use this instead of full ABC+XYZ+DOB.

Answer 3

This should give you the desired results:
select EKey = t.EKey ,
ABC = t.ABC ,
XYZ = t.XYZ ,
DOB = t.DOB ,
My_Min = t.My_Min ,
Row_Num = t.Row_Num ,
Required_Min = min(t.My_Min) over( partition by t.ABC , t.XYZ , t.DOB )
from dbo.my_table t
order by t.ABC ,
t.XYZ ,
t.DOB ,
t.Row_Num
If you actually want to add a new column to the table, then you'll need to do something like this:
alter table dbo.my_table add column Required_Min int null
go
update dbo.my_table
set Required_Min = s.Required_Min
from dbo.my_Table t
join ( select ABC , XYZ , DOB , Required_Min = min(My_Min)
from dbo.my_table
group by ABC , XYZ , DOB
) s on s.ABC = t.ABC
and s.XYZ = t.XYZ
and x.DOB = t.DOB
go
alter table dbo.my_table alter column Required_Min int not null
go
The update could also be standard SQL, using a correlated subquery instead of a derived table. Assuming decent indexing, the execution plans shouldn't vary too much:
update dbo.my_table
set Required_Min = ( select min(My_Min)
from dbo.my_table x
where x.ABC = dbo.my_table.ABC
and x.XYZ = dbo.my_table.XYZ
and x.DOB = dbo.my_table.DOB
)

Answer 4

I'm assuming ssms implies sql server. I'm assuming that you want the minimum "my_min" from rows with the same "RowNum"
try this:
update table_name as outer
set outer.Required_Min = (select min(My_Min)
from table_name as inner
where inner.RowNum = outer.RowNum)
-edit-
new assumption: ABC+XYZ+DOB form a natural key:
update table_name as outer
set outer.Required_Min = ( select min(My_Min)
from table_name as inner
where inner.ABC = outer.ABC
and inner.XYZ = outer.XYZ
and inner.DOB = outer.DOB )

Answer 5

Looks like a job for analitical func PARTITION BY:
select EKey, ABC, XYZ, DOB, My_Min, Row_Num,
min(My_Min) OVER (PARTITION BY ABC, XYZ, DOB) as Required_Min
from Stats
Also, I recommend you to do not store 'Required_Min' as a column, but compute it on demand in a view or query. Otherwise, you'll have to spend your time on providing data relevance (triggers or smth ). So, I would build specific inedexes instead to improve query perfomance.