Prev: Finding Predcied values with confidence level based on the regression model
Next: Jim and MMMMIIIIKKKKEEEE
From: laika on 9 Mar 2010 04:43
Hallo,

I can't find a sloution for my following problem.
With the next dataset :

ID Name Number
1 Obj 1025
1 Res 158
1 Num 1458
2 Obj 587
2 Res 21
3 Obj 900
3 Num 28

i want to create the following dataset :

ID Obj Res Num
1 1025 158 1458
2 1458 21 .
3 900 . 28

Is this possible? The Proc TRANSPOSE give me only the name of the
columns.

Tx
From: barryd on 9 Mar 2010 05:02
On Mar 9, 9:43 am, laika <michel.verhe...(a)axa.be> wrote:
> Hallo,
>
> I can't find a sloution for my following problem.
> With the next dataset :
>
> ID   Name   Number
> 1    Obj        1025
> 1    Res       158
> 1    Num      1458
> 2    Obj        587
> 2    Res       21
> 3    Obj        900
> 3    Num      28
>
> i want to create the following dataset :
>
> ID    Obj      Res    Num
> 1     1025    158     1458
> 2     1458      21          .
> 3       900        .         28
>
> Is this possible? The Proc TRANSPOSE give me only the name of the
> columns.
>
> Tx

Laika,

PROC TRANSPOSE seems ideal for this.

When I ran the code:

proc transpose data=a out=b;
by id;
id name;
run;

In this code your input dataset is a and the dataset you require is b.

I got exactly what you wanted - unless I've mis-understood your
requirement.

Perhaps you just missed out the ID statement.

In any event I hope this helps.

Regards,

BPD
From: ravipoddar.bit on 9 Mar 2010 06:57
On Mar 9, 2:43 pm, laika <michel.verhe...(a)axa.be> wrote:
> Hallo,
>
> I can't find a sloution for my following problem.
> With the next dataset :
>
> ID   Name   Number
> 1    Obj        1025
> 1    Res       158
> 1    Num      1458
> 2    Obj        587
> 2    Res       21
> 3    Obj        900
> 3    Num      28
>
> i want to create the following dataset :
>
> ID    Obj      Res    Num
> 1     1025    158     1458
> 2     1458      21          .
> 3       900        .         28
>
> Is this possible? The Proc TRANSPOSE give me only the name of the
> columns.
>
> Tx


This query will give u the desired output:


proc sql;

create table x as
select coalesce(p.id,z.id) as id,
obj,res,numb
from (select coalesce(x.id,y.id) as id,
obj,res
from (select id , number as Obj
from a
where name = 'Obj') as x
full outer join
(select id , number as Res
from a
where name = 'Res') as y
on (x.id = y.id)
) as p
full outer join
(select id , number as numb
from a
where name = 'Num') as z
on (p.id = z.id)
order by id;
quit;
 | 
Pages: 1
Prev: Finding Predcied values with confidence level based on the regression model
Next: Jim and MMMMIIIIKKKKEEEE