topic Re: Model target=1 instead of Target=0 in Nominal Logistic Regression in Discussions

Model target=1 instead of Target=0 in Nominal Logistic Regression

amitkb — Tue, 29 Jan 2013 14:42:28 GMT

Hi all,

In fitting a Nominal Logistics Regression, JMP models Target=0. I want to model target=1 and get the parametric estimates for the factors.

How can i do this.

I appreciate all your help.

Regards,

Amit Bothra

Re: Model target=1 instead of Target=0 in Nominal Logistic Regression

ancatilea — Tue, 29 Jan 2013 15:01:08 GMT

Hi.

I am not sure if this applies/works in JMP

but there is an option in the

PROC LOGISTIC data = your_data DESCENDING;

model y = x1 x2;

run;

Maybe this helps?

Re: Model target=1 instead of Target=0 in Nominal Logistic Regression

amitkb — Tue, 29 Jan 2013 15:20:23 GMT

Hi Anca,

Thanks for the reply. In SAS its easy and there are multiple ways.

In JMP its wizard driven, so trying to figure out.

Regards,

Amit Bothra

Re: Model target=1 instead of Target=0 in Nominal Logistic Regression

ancatilea — Tue, 29 Jan 2013 16:36:38 GMT

Maybe this helps...(as I am completely unfamiliar with JMP)

Here is an extract from :

Logistic Fit Platform Options

[...]

For binary responses, the log odds ratio for flipped response levels involves only changing the sign of the parameter, so you might want the reciprocal of the reported value to focus on the last response level instead of the first.

[...]

Does this help?

Re: Model target=1 instead of Target=0 in Nominal Logistic Regression

amitkb — Tue, 29 Jan 2013 17:13:39 GMT

Hi Anca,

I found the way to model target=1.

In your data table, right click on the Target column and under column properties there is an option Value Ordering. You need to put the value 1 before 0. This will do the trick.

I appreciate your effort in helping me.

Regards,

Amit