How to create multiple regresion models from a list

[Devir]

Greetings, this is my first time using this webpage and I would appreciate if you could help me out. I need to create several linear models like the following:

proc glm data=tabla.transpuesto_lluvia plots=none;
	model _9198 =_9446 _9452 _9200;
run;

_9198, _9446, etc. are variables from a database (tabla.transpuesto_lluvia). My problem is that I have over 20 variables to model and I was wondering if I could autome the process with this idea in mind:

proc glm data=tabla.transpuesto_lluvia plots=none;
	model v1 =v2 v3 v4;
run;

These are the combination of variables that I want to model, being the first row the one in the example:

data tabla.correlaciones;
input v1 $ v2 $ v3 $ v4 $;
cards;
_9198 _9446 _9452 _9200
_9952 _9773 _9766 _9717
_9489 _9487 _9477 _9898
_9717 _9766 _9726 _9773
_9822 _9814 _9815 _9862
_9773 _9952 _9766 _9717
_9446 _9452 _9198 _9814
_9913 _9878 _9916 _9874
_9908 _9910 _9898 _9874
_9452 _9946 _9814 _9198
_9898 _9896 _9862 _9874
_9766 _9773 _9952 _9717
_9726 _9717 _9766 _9773
_9815 _9814 _9452 _9200
_9896 _9898 _9487 _9862
_9874 _9898 _9862 _9910
_9910 _9908 _9874 _9878
_9814 _9815 _9452 _9200
_9862 _9898 _9874 _9896
_9477 _9487 _9489 _9898
_9878 _9913 _9916 _9874
_9200 _9452 _9814 _9815
_9916 _9878 _9913 _9874
_9487 _9489 _9477 _9898
;
run;

Any idea of how could I create multiple models thanks to this last combination of variables? Thank you for your time.

[PeterClemmensen]

@Devir Hi and welcome to the SAS Community 🙂

Sure, this is possible. What is your desired result here? HTML output or SAS data sets?

[Devir]

I need to make some predictions with these models, so I think Im going to need a dataset to work with.
Thank you for your welcome!

[PeterClemmensen · Accepted solution]

Ok. First off, let us create one PROC GLM step for each combination in the correlaciones data set. You can do this with Call Execute Logic like this.

Notice that I have commented out the Call Execute Statement, since I don't have your data. You can see the generated code in the c variable in the Callstack data set. Each model creates an output data set Model+N in the work library. 

This should get you started. Feel free to ask 🙂

data correlaciones;
input v1 $ v2 $ v3 $ v4 $;
cards;
_9198 _9446 _9452 _9200
_9952 _9773 _9766 _9717
_9489 _9487 _9477 _9898
_9717 _9766 _9726 _9773
_9822 _9814 _9815 _9862
_9773 _9952 _9766 _9717
_9446 _9452 _9198 _9814
_9913 _9878 _9916 _9874
_9908 _9910 _9898 _9874
_9452 _9946 _9814 _9198
_9898 _9896 _9862 _9874
_9766 _9773 _9952 _9717
_9726 _9717 _9766 _9773
_9815 _9814 _9452 _9200
_9896 _9898 _9487 _9862
_9874 _9898 _9862 _9910
_9910 _9908 _9874 _9878
_9814 _9815 _9452 _9200
_9862 _9898 _9874 _9896
_9477 _9487 _9489 _9898
_9878 _9913 _9916 _9874
_9200 _9452 _9814 _9815
_9916 _9878 _9913 _9874
_9487 _9489 _9477 _9898
;
run;

data callstack;
   set correlaciones;
   c = compbl(cat(
   "proc glm data = tabla.transpuesto_lluvia plots=none outstat = model", _N_, ";
      model ", v1, " = ", v2, " ", v3, " ", v4, ";
    run;"
    ));
   *call execute(c);
run;

[Devir]

Thanks for your answer! I'll try.

[PaigeMiller]

PROC GLM should end with a QUIT; and not a RUN;