Building models with SAS Enterprise Miner, SAS Factory Miner, SAS Visual Data Mining and Machine Learning or just with programming

Comparing Logistic Regression in SAS Enteprise Miner and SAS Enterprise Guide

Accepted Solution Solved
Reply
Highlighted
New Contributor
Posts: 4
Accepted Solution

Comparing Logistic Regression in SAS Enteprise Miner and SAS Enterprise Guide

I am running a logistic regression to predict a binary outcome with 12 predictors.  I ran the logistic regression in Enterprise Guide and then in Enterprise Miner...all fit statistics match, but the intercept and some of the parameter estimates do not match. While some odds ratios in one output can be inverted to match the odds-ratios in the other, I cannot come up with a logical reason of why this is happening. In each I am using the binary outcome=1 as the level to fit model.  I attached output for reference.  Any help is appreciated!

 

 


Accepted Solutions
Solution
2 weeks ago
New Contributor
Posts: 4

Re: Comparing Logistic Regression in SAS Enteprise Miner and SAS Enterprise Guide

Reeza,

Thanks for your quick reply! When I program in SAS I do use(param=ref ref=first). I am new to EG and thought that because I had binary categorical variables I could 'drag' all variables (binary and quantitative) to the 'Quantitative' list and be ok... evidently dragging binary predictors to the 'categorical' list fixes the problem.

Thanks!!!

View solution in original post


All Replies
Super User
Posts: 23,776

Re: Comparing Logistic Regression in SAS Enteprise Miner and SAS Enterprise Guide

Are the parameterization methods the same for your categorical variables?

The default is GLM though REF is usually what people want.

 


@jnunner wrote:

I am running a logistic regression to predict a binary outcome with 12 predictors.  I ran the logistic regression in Enterprise Guide and then in Enterprise Miner...all fit statistics match, but the intercept and some of the parameter estimates do not match. While some odds ratios in one output can be inverted to match the odds-ratios in the other, I cannot come up with a logical reason of why this is happening. In each I am using the binary outcome=1 as the level to fit model.  I attached output for reference.  Any help is appreciated!

 

 


 

Solution
2 weeks ago
New Contributor
Posts: 4

Re: Comparing Logistic Regression in SAS Enteprise Miner and SAS Enterprise Guide

Reeza,

Thanks for your quick reply! When I program in SAS I do use(param=ref ref=first). I am new to EG and thought that because I had binary categorical variables I could 'drag' all variables (binary and quantitative) to the 'Quantitative' list and be ok... evidently dragging binary predictors to the 'categorical' list fixes the problem.

Thanks!!!
Super User
Posts: 23,776

Re: Comparing Logistic Regression in SAS Enteprise Miner and SAS Enterprise Guide

Glad to hear it's fixed, please mark this as solved. Marking your own answer as the solution here would also be appropriate. 

Cheers!

☑ This topic is solved.

Need further help from the community? Please ask a new question.

Discussion stats
  • 3 replies
  • 132 views
  • 2 likes
  • 2 in conversation