topic How to get R-square for individual time-series group in SAS Procedures

How to get R-square for individual time-series group

amanda_cz — Mon, 23 Feb 2015 06:43:17 GMT

Hi, I am running a time series model with multiple firms with multiple years of observation (more than 10,000 observations). I want to get R-squares for individual firms. I know that r-square will show up in the individual firms' output report.But I don't know how to obtain a data file includes both firm_id and also R-square for individual firms. Any suggestion will be appreciated.

my basic code is

PROC AUTOREG DATA=TS;

by firm_id;

MODEL price=bps;

OUTPUT OUT=OUT1;

RUN;

Original data sample:

firm_ID	year	price	bps
001	2004	0.15	0.02
001	2005	0.16	0.05
001	2006	0.13	0.03
001	2007	0.18	0.04
001	2008	0.19	0.06
002	2004	1.2	0.1
002	2005	1.15	0.3
002	2005	1.83	0.5
008	2001	3.5	0.33
008	2002	3.8	0.54
008	2003	3.6	0.63
008	2013	4.5	0.49
011	2004	3.1	4
011	2003	3.5	5
011	2005	3.7	3.5
011	2012	3.8	4.5

Re: How to get R-square for individual time-series group

SteveDenham — Mon, 23 Feb 2015 16:33:51 GMT

Try using ODS to get a dataset with the R squared values:

PROC AUTOREG DATA=TS;

by firm_id;

MODEL price=bps;

OUTPUT OUT=OUT1;

ods output fitstatistics=fitstatistics;

RUN;

This will generate a dataset in your WORK library with the regression R squared values for each BY variable.

Steve Denham

Re: How to get R-square for individual time-series group

amanda_cz — Mon, 23 Feb 2015 19:30:34 GMT

Hi, I tried to use this method. However, sas gives me an warning message and there seems to be an error in the code.

WARNING: Output 'fitstatistics' 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: The data set WORK.OUT1 has 6240 observations and 11 variables.

NOTE: PROCEDURE AUTOREG used (Total process time):

real time 0.04 seconds

cpu time 0.03 seconds

Re: How to get R-square for individual time-series group

SteveDenham — Mon, 23 Feb 2015 19:39:30 GMT

Since I was able to make this work by looping the example in the PROC AUTOREG documentation, I will guess that the ODS file name may be different in whatever version of SAS/ETS you are using. In order to check this out, wrap your proc autoreg block in ODS TRACE ON and ODS TRACE OFF and check the log for the table names.

Here is my example:

title1 'Lack of Fit Study';
title2 'Fitting White Noise Plus Autoregressive Errors to a Sine Wave';

data a;
   pi=3.14159;
   do grp = 1 to 3;
   do time = 1 to 75;
      if time > 75 then y = .;
      else y = sin( pi * ( time / 50 ) );
      x = ranuni( 1234567 );
      output;
   end;
   end;
run;
ods trace on;
proc autoreg data=a plots;
by grp;
   model y = x ;
   output out=b p=pred pm=xbeta;
   ods output fitsummary=fitsummary;
run;
ods trace off;

Steve Denham

Message was edited by: Steve Denham

Re: How to get R-square for individual time-series group

amanda_cz — Mon, 23 Feb 2015 19:57:12 GMT

Thanks so much for your help.

I switched from

ods output fitstatistics=fitstatistics;

ods output fitsummary=fitsummary;

Now it works.

Thanks si much for your help! I am grateful for it.