topic Re: Regression Output in SAS Programming

Regression Output

floraminty — Wed, 21 Jun 2017 15:08:21 GMT

Dear All,

In my course of statistics we are using SAS Software.

One of the exercises is :compute the fitted sales of good for a supermarket that pays 800 euros in promotion , that has a parking, has 800 000 euros as revenues and the price of the goods is 95 cents;

Using the previous data, i calculated:

Y= 0.555+ 2.042*x1+1.172*x2+1.467*x3-0.856*x4

Y= 0.555+ 2.042*800+1.172*1+1.467*800000-0.856*0.95

Y=1 175 234.5

Next task is:the regression output has been produced using SAS software. Write the programm to computethis output.

Could you please help me with this task?

Re: Regression Output

PaigeMiller — Wed, 21 Jun 2017 15:25:38 GMT

You can look up the documentation for PROC REG and give it a try yourself. There are examples in the documentation.

http://documentation.sas.com/?cdcId=statcdc&cdcVersion=14.2&docsetId=statug&docsetTarget=statug_reg_toc.htm&locale=en

Re: Regression Output

Reeza — Wed, 21 Jun 2017 16:03:03 GMT

Here's a list (and code) of the different ways to score a regression model.

Re: Regression Output

floraminty — Wed, 21 Jun 2017 16:08:31 GMT

Where?

Re: Regression Output

Reeza — Wed, 21 Jun 2017 16:47:27 GMT

http://blogs.sas.com/content/iml/2014/02/19/scoring-a-regression-model-in-sas.html

My bad, CTRL + V doesn't always work on this machine 🙂

Re: Regression Output

Rick_SAS — Wed, 21 Jun 2017 16:52:10 GMT

Do you actually have data? PROC REG fits a model to data. If you don't have data, then you can just write a DATA step that scores the model.

Re: Regression Output

PaigeMiller — Wed, 21 Jun 2017 16:56:07 GMT

@Rick_SAS wrote:

Do you actually have data? PROC REG fits a model to data. If you don't have data, then you can just write a DATA step that scores the model.

Sure you can write the code without data, but you can't run it. 😉

Re: Regression Output

Rick_SAS — Wed, 21 Jun 2017 17:02:13 GMT

I see the smiley face, but to clarify for the OP, I use the term "data" to mean observed values that are used to FIT the model. You can EVALUATE the model at any values of the explanatory variables. The values do not need to be observed. For your example, the DATA step is

data Want;
input X1-X4;
Y= 0.555+ 2.042*x1+1.172*x2+1.467*x3-0.856*x4;
datalines;
800 1 800000 0.95
;
proc print; run;

Re: Regression Output

PaigeMiller — Wed, 21 Jun 2017 17:06:23 GMT

Oh, I see, somehow the thread has morphed into scoring (or evaluating) an existing regression model; but that was not the original question. The original question was how to write SAS code to produce the output (regression coefficients and/or a regression model) that the user has provided. The PROC REG code to create these coefficients and estimate the model can of course be written without the data.

Re: Regression Output

Reeza — Wed, 21 Jun 2017 17:09:42 GMT

@PaigeMiller You're correct, the question is ambiguous.

The OP can clarify I assume what h/she needs.