Date:         Thu, 29 May 2008 10:10:57 +0530
Reply-To:     ajay ohri <ohri2007@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         ajay ohri <ohri2007@GMAIL.COM>
Subject:      Re: Dummy Values for categorical variables
Comments: To: "data _null_," <datanull@gmail.com>
In-Reply-To:  <4bc14e460805282120j403fc440xcc37c76e8e8231dd@mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1

Hi Data_Null,

I replaced missing values by MISS, and I dropped the extra variables. , and
it works .
 Thanks a ton!! Saves me a lot of time.

Regards,

Ajay

On Thu, May 29, 2008 at 9:50 AM, ajay ohri <ohri2007@gmail.com> wrote:

> Hi data_null,
>
> I am getting this error.
>
>
> ERROR: Integer divide by zero.
> ERROR: Termination due to Floating Point Exception
>
> I do have some non categorical variables also (numerical variables) , some
> categorical variables have values of a character or are set to missing ,
> some variables have numerical value 0. What do I need to do rectify this ?
>
> regards,
>
> ajay
>
>
> On Thu, May 29, 2008 at 8:53 AM, data _null_, <datanull@gmail.com> wrote:
>
>> data ajay;
>>   input (v70 v71)(:$3.) @@;
>>   cards;
>> vx1 ap1 rd3 dr2 uf sf
>> ;;;;
>>   run;
>> proc transreg;
>>   model class(_all_ / zero=none);
>>   output design;
>>   run;
>> proc print;
>>   run;
>>
>> On 5/28/08, ajay ohri <ohri2007@gmail.com> wrote:
>> > Dear List,
>> >
>> > I have some 70 variables that take values of different categories like
>> VAR
>> > 71 takes values of ap1,dr2 ,sf  etc where these values refer to
>> different
>> > categories.
>> >
>> > Is there a macro by which I can create dummay variables so that for each
>> > record I can create
>> >
>> >  var71_ap1 = 1 if var71 =ap1 else var71_ap1 = 0 ;
>> > var71_dr2 = 1 if var71 =dr2 else var71_dr2 = 0 ;
>> > var71_sf = 1 if var71 =sf else var71_sf = 0 ;
>> >
>> > What i am trying to do is to see which permutation and combination of
>> > variables (i.e var 2=ab, var71=ap1 ) gives me the maximum of  Var1 ,if
>> at
>> > all ( where var 1 is a continuous numerical variable ). Is this the
>> right
>> > approach ? To add to the fun the total number of records is very low ,
>> > something like 1000 records. I am trying to rank order the top 10
>> > combinations of variables that are significant  in number of records as
>> well
>> > as in contributing the most to the dependent variable.
>> >
>> > Could anyone help me on this ?
>> >
>> > Regards,
>> >
>> > Ajay
>> > www.decisionstats.com
>> >
>>
>
>