BookmarkSubscribeRSS Feed
MUABer
Fluorite | Level 6

We are using gampl on a very large data set. It runs fine but no confidence intervals. We tried gam but it was too slow to run the models.

We must, if possible, find a way to confer the beta estimates in risk ratios and calculate confidence intervals using gampl. Does any one have a suggestion?

Would using gam in R work better for this?

 

We asked SAS Support and got the answer below.

"L'Beta would be a contrast of the model parameters. This is provided in parametric modeling procedures like REG, GLM, GENMOD, and others via the CONTRAST and ESTIMATE statements. Since GAMPL models do not have a pre-defined set of parameters, confidence intervals for such parameters and contrasts of parameters are not available." David Schlotzhauer

 

Thank you.

1 REPLY 1
H
Pyrite | Level 9 H
Pyrite | Level 9

Hmm,

 

I was getting ready to post a question and came across this post. Were you able to resolve this?  I wonder if percentile bootstrap confidence intervals would be appropriate here. How did you get relative risks out of the model, log link?, and were you trying to get some type of relative risk for the spline term as well. I am trying to  find a good way to get estimates out of the spline term.

 

Thanks.

Ready to join fellow brilliant minds for the SAS Hackathon?

Build your skills. Make connections. Enjoy creative freedom. Maybe change the world. Registration is now open through August 30th. Visit the SAS Hackathon homepage.

Register today!
What is ANOVA?

ANOVA, or Analysis Of Variance, is used to compare the averages or means of two or more populations to better understand how they differ. Watch this tutorial for more.

Find more tutorials on the SAS Users YouTube channel.

Discussion stats
  • 1 reply
  • 1239 views
  • 0 likes
  • 2 in conversation