Min except for zero

We have a complex query with many aggregate functions.
For one column where we return MIN, we would like to return the minimum
non-zero value.
I don't think using HAVING is going to work, because we don't want to
exclude any rows from the result set.
We're thinking of doing something like MIN(case...) but haven't found
something
that works.
An example:
TAB1
NUM1 NUM2
3 0
5 1
7 2
We want to select MAX(NUM1) and MIN(NUM2) and get 7 and 1 as the result.Try
MIN(CASE WHEN NUM2 <> 0 THEN NUM2 END)
or
MIN(NULLIF(NUM2,0))
Steve Kass
Drew University
jhcorey@.yahoo.com wrote:

>We have a complex query with many aggregate functions.
>For one column where we return MIN, we would like to return the minimum
>non-zero value.
>I don't think using HAVING is going to work, because we don't want to
>exclude any rows from the result set.
>We're thinking of doing something like MIN(case...) but haven't found
>something
>that works.
>An example:
>TAB1
>NUM1 NUM2
>3 0
>5 1
>7 2
>
>We want to select MAX(NUM1) and MIN(NUM2) and get 7 and 1 as the result.
>
>|||One way is to use CASE to convert the 0s to NULL:
SELECT MAX(a), MIN(CASE b WHEN 0 THEN NULL ELSE b END) AS b
FROM
(
SELECT 3 AS a, 0 AS b
UNION
SELECT 5, 1
UNION
SELECT 7, 2
) AS i
Tibor Karaszi, SQL Server MVP
http://www.karaszi.com/sqlserver/default.asp
http://www.solidqualitylearning.com/
<jhcorey@.yahoo.com> wrote in message news:1108483393.537431.133510@.c13g2000cwb.googlegroups
.com...
> We have a complex query with many aggregate functions.
> For one column where we return MIN, we would like to return the minimum
> non-zero value.
> I don't think using HAVING is going to work, because we don't want to
> exclude any rows from the result set.
> We're thinking of doing something like MIN(case...) but haven't found
> something
> that works.
> An example:
> TAB1
> NUM1 NUM2
> 3 0
> 5 1
> 7 2
>
> We want to select MAX(NUM1) and MIN(NUM2) and get 7 and 1 as the result.
>|||Correction,
select max(num1), min(case when num2 > 0 then num2 else null end) from tab1
go
AMB
"Alejandro Mesa" wrote:
> You are in the correct path. Try:
> select max(num1), min(case when num2 > 2 then num2 else null end) from tab
1
> go
>
> AMB
> "jhcorey@.yahoo.com" wrote:
>|||You are in the correct path. Try:
select max(num1), min(case when num2 > 2 then num2 else null end) from tab1
go
AMB
"jhcorey@.yahoo.com" wrote:

> We have a complex query with many aggregate functions.
> For one column where we return MIN, we would like to return the minimum
> non-zero value.
> I don't think using HAVING is going to work, because we don't want to
> exclude any rows from the result set.
> We're thinking of doing something like MIN(case...) but haven't found
> something
> that works.
> An example:
> TAB1
> NUM1 NUM2
> 3 0
> 5 1
> 7 2
>
> We want to select MAX(NUM1) and MIN(NUM2) and get 7 and 1 as the result.
>|||Hi
WHERE (Num1 > 0 OR Num2 > 0)
Regards
Mike
"jhcorey@.yahoo.com" wrote:

> We have a complex query with many aggregate functions.
> For one column where we return MIN, we would like to return the minimum
> non-zero value.
> I don't think using HAVING is going to work, because we don't want to
> exclude any rows from the result set.
> We're thinking of doing something like MIN(case...) but haven't found
> something
> that works.
> An example:
> TAB1
> NUM1 NUM2
> 3 0
> 5 1
> 7 2
>
> We want to select MAX(NUM1) and MIN(NUM2) and get 7 and 1 as the result.
>|||MIN(NULLIF(num2,0))
David Portas
SQL Server MVP
--|||Whew, that was quick!
Thanks to all.
As a note, I was trying this in the Northwind database and
having some difficulty (all values are decimals between 0 and 1):
select max(discount),min(discount),min(case when discount = 0 then null
else discount end) from [order details]
Jim

MIN and MAX strange results

I want to get the MIN and MAX value of a table column from a specified period of time. I execute a query and it return the result. The problem is that the values returned by MIN and MAX are not always correct!!

This is the result table

Date Statement From To

1 2007-01-03 00:00:00 Invoice 1 2 Correct
2 2007-01-04 00:00:00 Receipt 1 1 Correct
3 2007-01-04 00:00:00 Invoice 10 9 Wrong
4 2007-01-05 00:00:00 Receipt 2 5 Correct
5 2007-01-05 00:00:00 Invoice 100 99 Wrong
6 2007-01-08 00:00:00 Invoice 124 175 Correct
7 2007-01-09 00:00:00 Invoice 176 224 Correct
8 2007-01-10 00:00:00 Invoice 225 265 Correct

From =From Statement Number

To= To statement Number

The odd behavior happens when the number of digits changes. If the range of the column is 1 digit ie from 0 to 9 the values reported are ok. If the digits change then there is a problem as in line 3 and 5.

Any ideas why this odd behavior happens?

rectis:

I think you need to provide (1) the SQL Statement that is not working correctly and the definition of the table (or at least the relevant columns). My knee-jerk guess would be that you are coming to grief because your "From" and "To" fields are defined as varchar instead of numeric (or integer).

If in fact your "from" and "to" fields are defined as varchar you first need to make a determination to the usage of these fields -- that is see if the definition needs to be modified such that columns are reformatted into numeric (or integer) columns. You may need to compute your max as MIN(CONVERT(INTEGER, FROM)) and MAX(CONVERT(INTEGER,TO))

|||You are right. Thank you very much. I think my brain was stopped.The field was defined as varchar. Now the values are ok!

Mimic DBCC Memusage

Does anyone have a routine that returns results like DBCC MEMUSAGE used to return in older versions of SQL Server?

I think sys.dm_os_buffer_descriptors gets you close. Books Online shows a query that assembles some of the data you are looking for. I've modified this to be a bit more user-friendly.

Hope this helps,
Bryan

Code Snippet

SELECT

x.name,

obj.name as [object_name],

index_id,

count(*)AS cached_pages_count,

is_modified

FROMsys.dm_os_buffer_descriptorsAS bd

INNERJOIN

(

SELECTobject_name(object_id)ASname

,index_id ,allocation_unit_id

FROMsys.allocation_unitsAS au

INNERJOINsys.partitionsAS p

ON au.container_id = p.hobt_id

AND(au.type = 1 OR au.type = 3)

UNION ALL

SELECTobject_name(object_id)ASname

,index_id, allocation_unit_id

FROMsys.allocation_unitsAS au

INNERJOINsys.partitionsAS p

ON au.container_id = p.hobt_id

AND au.type = 2

)AS obj

ON bd.allocation_unit_id = obj.allocation_unit_id

innerjoinsys.databases x

on bd.database_id=x.database_id

GROUPBY x.name, obj.name,index_id, is_modified

ORDERBY 1, 2, 3, 5

|||

Thanks, Bryan. This is similar to what I presently have, but it might fix the problems; hang on and I'll give this a try.

This looks pretty good; thank you.