Date:         Thu, 11 Nov 2004 12:49:34 -0500
Reply-To:     friendly@yorku.ca
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Michael Friendly <friendly@YORKU.CA>
Organization: York University
Subject:      Re: help with Box-Cox
Content-Type: text/plain; charset=us-ascii; format=flowed

No surprise!  X**0 = 1!

lambda=0 in the Box-Cox transformation means log(y), so you need to
code something like

if lambda=0 then result=log(x);
else result=x**lambda/lambda;

I have a boxcox macro (for proc reg) and a boxglm macro (for glm)
at
http://www.math.yorku.ca/SCS/sasmac/

I'm not sure why you are using proc model for OLS, except to
include lambda as a parameter, but that is easily accomplished
with %boxcox using REG

-Michael

gabe wrote:
> Hi,
>
> I'm trying to estimate the equation below using a Box-Cox
> transformation on the variable uc_g_M1. I'm trying to test lambda = 0
> & lambda =1. However, for values of lambda = 0 the model collapses. Is
> there another way to estimate this model with a Box-Cox
> transformation?
>
> proc model data=calculations;
>       var log_real_c log_real_g_M1 uc_g_M1;
>       parms ln_a gamma cons_elasticity lambda;
>       log_real_g_M1 = log_a + cons_elasticity*log_real_c -
> gamma*((uc_g_M1**lambda)/lambda);
> fit log_real_g_M1 /ols outest=coef_M1_MD;
> run;
> quit;
>
> Thank you,
>
> gabriel


--
Michael Friendly     Email: friendly@yorku.ca
Professor, Psychology Dept.
York University      Voice: 416 736-5115 x66249 Fax: 416 736-5814
4700 Keele Street    http://www.math.yorku.ca/SCS/friendly.html
Toronto, ONT  M3J 1P3 CANADA