topic Re: proc gampl logistic regression splines adjusted model in Statistical Procedures

proc gampl logistic regression splines adjusted model

ChristosK — Fri, 09 Feb 2018 00:31:39 GMT

Having trouble with creating a fully adjusted logistic model using proc gampl at this stage:

We were trying to get a smoothed out spline figure for the adjusted model...

This is the error message:

NOTE: The SAS System stopped processing this step because of errors.

75 id Ind_DBP1 Vent__240;

76 model Vent__240(reference ='0') = spline(Ind_DBP1) spline(age)

77 Sex_num Redo1 NIDDM1 COPD Hyperlip CHF Afib CM CAD MI AICD DDRF PVD BB Steroids Amio AA1 Statin1 Nitrates1 Diurectics1

77 ! IABP_ever1

ERROR 22-322: Syntax error, expecting one of the following: ;, /, PARAM, SPLINE.

ERROR 76-322: Syntax error, statement will be ignored.

78 Age pre_Na pre_BUN pre_Cr pre_Gluc pre_PTT pre_Neutr o pre_Lymph pre_Hgb

79 / dist=binary;

80 output out = spline3age upper = u lower = l predicted = fit;

81 run;

83 OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;

ERROR: Calling global tagset function 'detach_stream':

Out of space writing to file /tmp/SAS_workB8270000F42E_localhost.localdomain/#LN01123.

title "Age as covariate";
proc data = forgampl3;
id Ind_DBP1 Vent__240;
model Vent__240(reference ='0') = spline(Ind_DBP1) spline(age)
Sex_num Redo1 NIDDM1 COPD Hyperlip CHF Afib CM CAD MI AICD DDRF PVD BB Steroids Amio AA1 Statin1 Nitrates1 Diurectics1 IABP_ever1
Age pre_Na pre_BUN pre_Cr pre_Gluc pre_PTT pre_Neutr o pre_Lymph pre_Hgb
/ dist=binary;
output out = spline3age upper = u lower = l predicted = fit;
run;

Re: proc gampl logistic regression splines adjusted model

ChristosK — Fri, 09 Feb 2018 00:49:24 GMT

The only thing I just tried was adding param ( ) to the covariates in the proc gampl statement.

Any thoughts?

param (
Redo1 NIDDM1 COPD Hyperlip CHF Afib CM CAD MI AICD DDRF PVD BB Steroids Amio AA1 Statin1 Nitrates1 Diurectics1 IABP_ever1
/* Age pre_Na pre_BUN pre_Cr pre_Gluc pre_PTT pre_Neutro pre_Lymph pre_Hgb */
)

Re: proc gampl logistic regression splines adjusted model

StatDave — Fri, 09 Feb 2018 13:53:16 GMT

All of your predictors must be either in PARAM( ) or in SPLINE( ). None can be listed outside of one of those. Also, I assume your code actually says "proc gampl data=forgampl3" and that you didn't leave out "gampl".

Re: proc gampl logistic regression splines adjusted model

ChristosK — Fri, 09 Feb 2018 17:19:48 GMT

Thanks Steve.

Yes it does. I had uploaded the code I had used in my original question but do not see it there now.

That happens alot when I start a new thread. It takes several attempts to get it to actually post.

Much appreciated.

Christos