topic Re: Is there any way I can apply 'solution' for just some certain variables while I use glm procedur in Statistical Procedures

Is there any way I can apply 'solution' for just some certain variables while I use glm procedure?

MonoLight — Wed, 15 May 2019 10:14:08 GMT

Hi! I have several variables with different scale types for independent variables.

There are variables with nominal scales, so I tried to use glm procedure with solution in model. like,

Proc glm data=pilot.pilot3;
class Purpose Categories Twitter_Posts Twitter_Retweets News_Posts;
model Points = Purpose Categories Twitter_Posts Twitter_Retweets News_Posts / solution;

But as you noticed in that code, I have ratio variables like 'Twitter Posts', 'Twitter_Retweets', and 'News_Posts'. You can expect to see hundreds of dummy variables will be created when I use solution in model.

So... I want to ask that if there is any way to apply solution in just Purpose and Categories, while it doesn't affect other ratio variables.

Or if you have any better idea to analyze an influence of those variables, please feel free to help me 😄

Thanks in advance!

Re: Is there any way I can apply 'solution' for just some certain variables while I use glm procedur

FreelanceReinh — Wed, 15 May 2019 12:21:12 GMT

Hi @MonoLight,

@MonoLight wrote:

I have ratio variables like 'Twitter Posts', 'Twitter_Retweets', and 'News_Posts'.

Just remove these from the CLASS statement, which is for categorical variables only. (Cf. Example 48.4 in the PROC GLM documentation).

Re: Is there any way I can apply 'solution' for just some certain variables while I use glm procedur

MonoLight — Thu, 16 May 2019 04:34:04 GMT

Ahha!!! Thanks! I thought CLASS should include every independent variables! Thanks again!