topic Re: PROC TRANSREG AND ANOVA in Statistical Procedures

PROC TRANSREG AND ANOVA

ANKH1 — Tue, 13 Aug 2019 16:26:00 GMT

What would be the procedure equivalent to proc transreg for ANOVAS (categorical IV's)? We are looking to use boxcox transformations on a continuous DV.

Re: PROC TRANSREG AND ANOVA [how to improve your question]

Community_Guide — Tue, 13 Aug 2019 17:58:20 GMT

Hello @ANKH1,

Your question requires more details before experts can help. Can you revise your question to include more information?

Review this checklist:

Specify a meaningful subject line for your topic. Avoid generic subjects like "need help," "SAS query," or "urgent."
When appropriate, provide sample data in text or DATA step format. See this article for one method you can use.
If you're encountering an error in SAS, include the SAS log or a screenshot of the error condition. Use the Photos button to include the image in your message.
It also helps to include an example (table or picture) of the result that you're trying to achieve.

To edit your original message, select the "blue gear" icon at the top of the message and select Edit Message. From there you can adjust the title and add more details to the body of the message. Or, simply reply to this message with any additional information you can supply.

SAS experts are eager to help -- help them by providing as much detail as you can.

This prewritten response was triggered for you by fellow SAS Support Communities member @PGStats

Re: PROC TRANSREG AND ANOVA

PaigeMiller — Tue, 13 Aug 2019 18:03:02 GMT

@ANKH1 wrote:

What would be the procedure equivalent to proc transreg for ANOVAS (categorical IV's)? We are looking to use boxcox transformations on a continuous DV.

You would want to use the CLASS transformation of the MODEL statement.

Re: PROC TRANSREG AND ANOVA

ANKH1 — Tue, 13 Aug 2019 18:08:27 GMT

Hi,
I tried the following:
74 proc transreg data = sample1;
75 model boxcox(DV)=class(CAT);
76 run;

ERROR: 11 invalid values were encountered while attempting to transform variable DV.

The 11 invalid values are from values=0. Moreover, when I write

proc transreg data = sample1;
model boxcox(DV/lambda=1)=class(IV);
run;

It shows exactly the same error.

Re: PROC TRANSREG AND ANOVA

PaigeMiller — Tue, 13 Aug 2019 18:51:00 GMT

Are you sure that the error is not because you have negative values of DV?

Please show us a portion of your data set SAMPLE1, following these instructions: https://communities.sas.com/t5/SAS-Communities-Library/How-to-create-a-data-step-version-of-your-data-AKA-generate/ta-p/258712

Re: PROC TRANSREG AND ANOVA

ANKH1 — Tue, 13 Aug 2019 19:20:10 GMT

Hi,

Thank you for your answer.

We don't have negative values in that DV. I tried following the code you provided to show our data but I guess I couldn't figured out how since our data set is an imported excel file with approx 19900 rows and 80 columns.

I tried the code with another DV to see if it could be another reason, but again it showed an error for one value. I checked for that DV and that one value seems to be again coming from only one value = 0.

Re: PROC TRANSREG AND ANOVA

PaigeMiller — Tue, 13 Aug 2019 19:25:12 GMT

As an experiment, run the PROC TRANSREG on a data set which doesn't have 0 values for Y using

model boxcox(DV/lambda=1)=class(IV);

and see if the error goes away.

Re: PROC TRANSREG AND ANOVA

ANKH1 — Tue, 13 Aug 2019 19:29:13 GMT

It did run!

Re: PROC TRANSREG AND ANOVA

PaigeMiller — Tue, 13 Aug 2019 19:34:19 GMT

Okay, but I can't explain why 0 is considered an invalid value for Y in the Box-Cox transformation with λ=1. Maybe @PGStats or @ballardw has some insight here. Or you could always talk to SAS tech support.

Re: PROC TRANSREG AND ANOVA

ANKH1 — Tue, 13 Aug 2019 19:42:18 GMT

Thank you!

Re: PROC TRANSREG AND ANOVA

ballardw — Tue, 13 Aug 2019 23:05:09 GMT

@ANKH1 wrote:

Hi,

Thank you for your answer.

We don't have negative values in that DV. I tried following the code you provided to show our data but I guess I couldn't figured out how since our data set is an imported excel file with approx 19900 rows and 80 columns.

I tried the code with another DV to see if it could be another reason, but again it showed an error for one value. I checked for that DV and that one value seems to be again coming from only one value = 0.

How exactly did you check for negative values? Since you say that you "imported" and Excel file there are times when, depending on the display settings in Excel very small magnitude negative value, such as -0.000001, could appear as 0.00 with the default 2 decimals in Excel and Proc Import might honor that setting or just not show if the value has a BEST format associated.

Consider:

data example;
   x= -0.00001;
   format x best4.;
   put x= ;
run;

The log will show x=0. As would a table view or proc print.

This may be a long shot but "mystery" behaviors sometimes require examining the less common issues.

Re: PROC TRANSREG AND ANOVA

PaigeMiller — Wed, 14 Aug 2019 12:03:55 GMT

That's an interesting thought.

This implies that @ANKH1 ought to do something like

proc print data=sample1(where=(dv<0));

Re: PROC TRANSREG AND ANOVA

Rick_SAS — Wed, 14 Aug 2019 13:20:50 GMT

Or use PROC MEANS to examine the minimum:

proc means data=sample1 fw=18 macdec=16;
   var dv;
run;

Re: PROC TRANSREG AND ANOVA

ANKH1 — Wed, 14 Aug 2019 16:29:47 GMT

Hi,

I ran the code and didn't get any negative values.

Re: PROC TRANSREG AND ANOVA

PaigeMiller — Wed, 14 Aug 2019 17:37:00 GMT

So, that's not it, but I don't really have any other suggestions about what might be causing this. I would still contact SAS Tech support to see if they have an explanation.

Re: PROC TRANSREG AND ANOVA

ANKH1 — Wed, 14 Aug 2019 17:41:11 GMT

Thank you!

Re: PROC TRANSREG AND ANOVA

Rick_SAS — Wed, 14 Aug 2019 18:00:35 GMT

The documentation for the TRANSREG section on the Box-Cox transformation says,

"This family of transformations of the positive [emphasis added] dependent variable y is controlled by the parameter ...."

It also states that you can "specify the PARAMETER=c transformation option when you want to shift the values of y, usually to avoid negatives." So when DV=0, you can to get rid of the error and apply the BC transformation by adding any positve value, such as the following (which adds 1):

proc transreg data=Have;
   model BoxCox(DV / lambda=1 parameter=1) = class(cat);
run;

Re: PROC TRANSREG AND ANOVA

ANKH1 — Wed, 14 Aug 2019 18:43:27 GMT

Hi!

Thanks, by adding the parameter=1 it ran.

proc transreg details data=sample1 ;
model boxcox(NFSDDFI / lambda = -3 to 3 by 0.25 parameter =1) = CLASS(ANIMALCAT);
run;

I looked in "Details" and this was the output:

It recommends lambda=0. I log transformed but it didn't work since I have 11 zeros.

Below is the code:

DATA sample2;
SET sample1;
TRANSFI=log(NFSDDFI);
RUN;

proc glm data=sample2;
class ANIMALCAT;
model TRANSFI= ANIMALCAT;
output out = notrans r= resid;
run;
symbol1 i=sm70;

proc univariate data=notrans noprint;
var resid;
histogram resid/normal kernel;
qqplot resid/normal (mu = est sigma=est);
run;

I tried using another DV that had no zeros but when I transformed the DV with the recommended lambda the DV was not normalized.

Am I reading the output from proc transreg correctly?

Re: PROC TRANSREG AND ANOVA

Rick_SAS — Wed, 14 Aug 2019 19:19:31 GMT

Yes, you are reading the output correctly, but you need to incorporate the PARAMETER=1 information:

DATA sample2;
SET sample1;
TRANSFI=log(NFSDDFI + 1);
RUN;

Re: PROC TRANSREG AND ANOVA

ANKH1 — Wed, 14 Aug 2019 19:46:43 GMT

Perfect! But according to K-S the resids are still not normal, even though the histogram looks pretty normal to me. What should I do?