topic Re: how to run regression repeatedly? in New SAS User

how to run regression repeatedly?

srikanthyadav44 — Thu, 05 Sep 2019 05:02:16 GMT

dear all,

i have to run the regression where in the independent varaibles will be same, but dependent variables will change each time.

I have nearly 200 dependent variables.

i have to run the regression repeatedly for 200 times for each dependent variable.

Finally, I have to save the R-squared and adjusted R-squared value of each regression in a separate file along with the name of the dependent variable.

the format of my file is as follows

IV_1	IV_2	DV_1	DV_2	…………….	DV_200
-1.67137	-1.28567	-1.41424	-1.7678	…………….	-1.83851
-0.20913	-0.16087	-0.17696	-0.2212	…………….	-0.23005
-0.41298	-0.31768	-0.34944	-0.4368	…………….	-0.45428
-0.61455	-0.47273	-0.52001	-0.65001	…………….	-0.67601
0.260277	0.200213	0.220235	0.275293	…………….	0.286305
-0.13165	-0.10127	-0.1114	-0.13925	…………….	-0.14482
0.065317	0.050244	0.055268	0.069085	…………….	0.071849
2.116535	1.628104	1.790914	2.238643	…………….	2.328188
-0.64197	-0.49382	-0.54321	-0.67901	…………….	-0.70617
1.636049	1.258499	1.384349	1.730436	…………….	1.799654
-0.04114	-0.03165	-0.03481	-0.04351	…………….	-0.04525
-1.18318	-0.91014	-1.00116	-1.25144	…………….	-1.3015
0.87525	0.673269	0.740596	0.925745	…………….	0.962775
0.202968	0.156129	0.171742	0.214678	…………….	0.223265
0.516805	0.397543	0.437297	0.546621	…………….	0.568486
0.137332	0.10564	0.116204	0.145255	…………….	0.151065
-1.62652	-1.25117	-1.37628	-1.72035	…………….	-1.78917
-2.7442	-2.11093	-2.32202	-2.90252	…………….	-3.01862
-0.20355	-0.15658	-0.17224	-0.2153	…………….	-0.22391
-0.12738	-0.09799	-0.10779	-0.13473	…………….	-0.14012
-0.99255	-0.7635	-0.83985	-1.04981	…………….	-1.0918
0.33774	0.2598	0.28578	0.357225	…………….	0.371514

IV is independent variable and DV is dependent variable

please helpe me writing a SAS code for this

thanks in advance

Re: how to run regression repeatedly?

Kurt_Bremser — Thu, 05 Sep 2019 07:36:08 GMT

Transpose, sort and use by in proc reg. Then it all happens in one step.

Re: how to run regression repeatedly?

Kurt_Bremser — Thu, 05 Sep 2019 07:41:16 GMT

Just a quick example:

data have;
infile datalines dlm='09'x dsd;
input IV_1	IV_2	DV_1	DV_2;
datalines;
-1.67137	-1.28567	-1.41424	-1.7678
-0.20913	-0.16087	-0.17696	-0.2212
-0.41298	-0.31768	-0.34944	-0.4368
-0.61455	-0.47273	-0.52001	-0.65001
0.260277	0.200213	0.220235	0.275293
-0.13165	-0.10127	-0.1114	-0.13925
0.065317	0.050244	0.055268	0.069085
2.116535	1.628104	1.790914	2.238643
-0.64197	-0.49382	-0.54321	-0.67901
1.636049	1.258499	1.384349	1.730436
-0.04114	-0.03165	-0.03481	-0.04351
-1.18318	-0.91014	-1.00116	-1.25144
0.87525	0.673269	0.740596	0.925745	
0.202968	0.156129	0.171742	0.214678
0.516805	0.397543	0.437297	0.546621
0.137332	0.10564	0.116204	0.145255	
-1.62652	-1.25117	-1.37628	-1.72035
-2.7442	-2.11093	-2.32202	-2.90252	
-0.20355	-0.15658	-0.17224	-0.2153	
-0.12738	-0.09799	-0.10779	-0.13473
-0.99255	-0.7635	-0.83985	-1.04981
0.33774	0.2598	0.28578	0.357225
;

proc transpose data=have out=trans;
by iv_1 iv_2 notsorted;
var dv:;
run;

proc sort data=trans;
by _name_;
run;

Re: how to run regression repeatedly?

PaigeMiller — Thu, 05 Sep 2019 11:00:32 GMT

@srikanthyadav44 wrote:

dear all,

i have to run the regression where in the independent varaibles will be same, but dependent variables will change each time.

I have nearly 200 dependent variables.

i have to run the regression repeatedly for 200 times for each dependent variable.

Finally, I have to save the R-squared and adjusted R-squared value of each regression in a separate file along with the name of the dependent variable.

PROC REG allows multiple dependent variables. Thus you need one PROC REG step. No need to transpose your data.

Use

ods output modelfit=modelfit;

to capture the R-squared values.

Re: how to run regression repeatedly?

ballardw — Thu, 05 Sep 2019 14:33:30 GMT

@srikanthyadav44 wrote:

dear all,

i have to run the regression where in the independent varaibles will be same, but dependent variables will change each time.

I have nearly 200 dependent variables.

i have to run the regression repeatedly for 200 times for each dependent variable.

Finally, I have to save the R-squared and adjusted R-squared value of each regression in a separate file along with the name of the dependent variable.

Since there are quite a few procedures in SAS that perform regression perhaps you can share the code you would run for one of the dependent variables.

Some procedures require slightly different methods of placing output into data sets, so your request for the r-squared may rely on the procedure. And possibly which r-squared for some sorts of regressions.

And what is to be done with the 200 output data sets?? You may want to rethink why you think you need so many data sets. This sounds like your next steps my involve lots of "loopy" code that may not be needed.

Re: how to run regression repeatedly?

srikanthyadav44 — Fri, 06 Sep 2019 05:39:13 GMT

dear KurtBremser

how to modify the SAS code suggested by you, if i want to run the code with an excel file imported into SAS.

thanks in advance

Re: how to run regression repeatedly?

Kurt_Bremser — Fri, 06 Sep 2019 07:04:10 GMT

Just apply the transpose and sort, using the actual variable names.

Re: how to run regression repeatedly?

srikanthyadav44 — Fri, 06 Sep 2019 07:31:25 GMT

dear KurtBremser

thanks for your prompt reply.

I understood tranpose and sort, but what changes i have to do to the following code to run the regression with imported excel file.

I may be right that the below code is for the data directly entered in SAS, because you are using datalines.

data have;
infile datalines dlm='09'x dsd;
input IV_1	IV_2	DV_1	DV_2;
datalines;

and also can you please explain me what is 'dlm='09' x dsd

Re: how to run regression repeatedly?

PaigeMiller — Fri, 06 Sep 2019 12:02:21 GMT

Hello, @srikanthyadav44

A much simpler approach appears in message 4 of this thread. No transposing needed.

Re: how to run regression repeatedly?

Kurt_Bremser — Fri, 06 Sep 2019 15:22:11 GMT

This step is just there to create an example dataset, in a way that allows everybody who reads the post to recreate it with a simple copy/paste and submit. It is not needed for your application, as you already have your data.

Re: how to run regression repeatedly?

srikanthyadav44 — Thu, 26 Sep 2019 17:53:58 GMT

dear PaigeMiller

i am unable to solve the problem of running the regression repeatedly and saving the R-squared values in a separate sheet.

kindly suggest me a SAS code more clearly and completely.

thanks in advance

Re: how to run regression repeatedly?

PaigeMiller — Thu, 26 Sep 2019 20:10:31 GMT

proc reg data=have;
    model dv_1-dv_200 = iv_1 iv_2;
run;
quit;

Re: how to run regression repeatedly?

srikanthyadav44 — Fri, 27 Sep 2019 04:31:34 GMT

thank you Mr. PaigeMiller. for your prompt reply

the SAS code suggested by you is working excellently.

after getting the regression results i run the code suggested by you earlier to get the R-squared values in a separate file

"ods output modelfit=modelfit;"

But it is not working. my SAS university edition software is hanging when i run that code.

if there is any other alternative way to get the R-squared values, please suggest me.
thanks in advance

Re: how to run regression repeatedly?

PaigeMiller — Fri, 27 Sep 2019 10:45:01 GMT

But it is not working.

This is not enough information. You need to show us the LOG of PROC REG, including the code and error message. Please click on the {i} icon and paste the log into the window that appears. Do not skip this step.

Re: how to run regression repeatedly?

srikanthyadav44 — Mon, 30 Sep 2019 05:00:18 GMT

dear Mr. PaigeMiller

when i run the " ods output modelfit=modelfit;" i am gettting the following message in log

OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;

NOTE: ODS statements in the SAS Studio environment may disable some output features.

74 ods output modelfit=modelfit;

76 OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;

please help in resolving this problem

thanks in advance

Re: how to run regression repeatedly?

srikanthyadav44 — Mon, 30 Sep 2019 08:37:06 GMT

dear Mr. Paigemiller

i run the SAS code as following and then i got the error as mentioned in the previous reply

proc reg data=stocks;
model Y1 - Y5=X1 X2;
run; 
quit;
ods output modelfit=modelfit;

Re: how to run regression repeatedly?

PaigeMiller — Mon, 30 Sep 2019 10:51:12 GMT

ODS OUTPUT goes immediately before the PROC statement, or inside the PROC, before RUN;

Re: how to run regression repeatedly?

srikanthyadav44 — Mon, 30 Sep 2019 11:06:37 GMT

dear Mr. Paigemiller

as you have replied, i run the code in two different ways. Still i am not getting the required output

regression output is generated, but R-squared values are not produced separately.

i used the code as follows

ods output modelfit=modelfit;
proc reg data=stocks;
model 	Y1- Y5= X1 X2;
RUN;
QUIT;

OR 
proc reg data=stocks;
model Y1- Y5= X1 X2;
ods output modelfit=modelfit;
RUN;
QUIT;


 I am getting the following error in log 


OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;
NOTE: ODS statements in the SAS Studio environment may disable some output features.
73 
74 proc reg data=stocks;
75 model Y1- Y5= X1 X2;
76 ods output modelfit=modelfit;
77 RUN;

78 QUIT;

WARNING: Output 'modelfit' was not created. Make sure that the output object name, label, or path is spelled correctly. Also, 
verify that the appropriate procedure options are used to produce the requested output object. For example, verify that 
the NOPRINT option is not used.
NOTE: PROCEDURE REG used (Total process time):
real time 17.00 seconds
cpu time 3.96 seconds

kindly help me out

thanks in advance

Re: how to run regression repeatedly?

PaigeMiller — Mon, 30 Sep 2019 12:03:23 GMT

This appears to be my mistake.

It should say

ods output FitStatistics=FitStatistics;

Re: how to run regression repeatedly?

srikanthyadav44 — Tue, 01 Oct 2019 05:22:16 GMT

Dear Mr. Paigemiller

i run the code successfully.

thanks a lot for your help.