Proc GLM interaction term with only specific levels

Dedes93 — Thu, 07 Dec 2017 14:11:53 GMT

Hi,

I need to run a linear regression with two categorical variables with 3 plus levels and one continuous variable on a continuous response variable, and is Thus using proc GLM.

Continuous variables:

- Price

- rating

Categorical varibles:

- product=(agriculture, rental, industry)

- customer_segment=(key, focus, corporate)

Is it possible to run the following regression, and if so how do you specify the interaction terms:

Proc GLM data=model1_data2 alpha=0.75;

class product customer_segment;

model price=rating product customer_segment industry*rating industry*key / solution;

output out=model1_v2_out r=res p=benchmark2 lcl=Q25 ucl=Q75;

Data model1_data2;

set model1_v2_out;

run;

So my issue is that i don't know how to specify interaction terms only involving specific levels of the categorical variable.

Hopes this makes sence.

I'm using SAS 9.4

Best, Andreas

Re: Proc GLM interaction term with only specific levels

WarrenKuhfeld — Thu, 07 Dec 2017 14:51:25 GMT

https://blogs.sas.com/content/iml/2016/02/22/create-dummy-variables-in-sas.html

Many procedures output coded variables. You can then process the output data set to drop terms. Personally, I like transreg because 1) I wrote it, 2) it has nice names and labels for terms, and 3) it outputs a macro variable with all of the terms in the model.

topic Re: Proc GLM interaction term with only specific levels in SAS Procedures

Proc GLM interaction term with only specific levels

Re: Proc GLM interaction term with only specific levels