topic Re: error proc panel : in SAS Forecasting and Econometrics

error proc panel :

Am_fr — Fri, 23 Sep 2022 16:34:25 GMT

Hello,

I am currently working on daily returns of several different REITs and I have a problem when I run my panel regression, it keeps showing the same error message :

"ERROR: The data set THESIS.DIVERSIFIED_RETURNS is not sorted in ascending sequence with respect to the time series ID. The current time period has date=31DEC2014 and the previous time period has date=31DEC2014 in cross-section CONM=AGELLAN COMMERCIAL REIT."

here is my code :

proc sort data=thesis.diversified_returns out=thesis.diversified_returns;
by company date;
run;

proc panel data=thesis.diversified_returns;
id company date;
model log_return=log_growth_in_cases log_AT log_FFO log_CASSET log_MVASSET/
fixtwo;
run;

here is an overview of my database:

datalines;

31DEC2009 60101010 ONE LIBERTY PROPERTIES INC 9.1800 2010 -1.453636611 -2.679457736 0 6.012947135

31DEC2009 60101010 PS BUSINESS PARKS 49.1800 2010 -0.24792172 -2.016888923 0 7.3555273585

05JAN2010 60101010 ONE LIBERTY PROPERTIES INC 9.7900 2010 -1.453636611 -2.679457736 0 6.012947135

05JAN2010 60101010 PS BUSINESS PARKS 48.2900 2010 -0.24792172 -2.016888923 0 7.3555273585

06JAN2010 60101010 ONE LIBERTY PROPERTIES INC 10.0400 2010 -1.453636611 -2.679457736 0 6.012947135

Thank you !

Re: error proc panel : ERROR The data set is not sorted in ascending sequence

PaigeMiller — Fri, 23 Sep 2022 13:54:40 GMT

From now on, please, when an ERROR appears in the log, show us the ENTIRE log for that PROC or DATA step, not just a few lines.

The current time period has date=31DEC2014 and the previous time period has date=31DEC2014

I'm not really a heavy user of PROC PANEL, but I don't think two observations with the exact same date are allowed.

Re: error proc panel :

Ksharp — Sat, 24 Sep 2022 10:36:23 GMT

Plz post it at Forecasting Forum:
PROC PANEL is under SAS/ETS
https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/bd-p/forecasting_econometrics

Re: error proc panel :

sbxkoenk — Sat, 24 Sep 2022 10:53:04 GMT

Hello,

You have duplicate dates within the same cross-section.

Might be a data quality problem.

If not, you can de-duplicate by taking the median (or whatever summary statistic) per cross-section (like company , firm, country , subject_id , ...) for the same dates.

You can do the latter using :

PROC TIMESERIES or
PROC TIMEDATA (or timeData cas action)

It's pretty easy, but let us know if you need an example !

Cheers,

Koen

Re: error proc panel :

Am_fr — Sat, 24 Sep 2022 12:57:55 GMT

thank you for your answer !!

but i believe I really don't have duplicate dates within the same cross-section.

when It is ordered by company name and date, the first lines of my table are these ones (where i don't see any duplicate) :

datalines;

017187 31JAN2014 60101010 AGELLAN COMMERCIAL REIT 7.7170 2014 -1.009485011

017187 03FEB2014 60101010 AGELLAN COMMERCIAL REIT 7.8444 2014 -1.009485011

017187 11FEB2014 60101010 AGELLAN COMMERCIAL REIT 8.5922 2014 -1.009485011

017187 23JUL2014 60101010 AGELLAN COMMERCIAL REIT 8.6800 2014 -1.009485011

017187 24JUL2014 60101010 AGELLAN COMMERCIAL REIT 8.6790 2014 -1.009485011

and it still returns the error message :

ERROR: The data set THESIS.DIVERSIFIED_RETURNS is not sorted in ascending sequence with respect to the time series ID. The current time period has date=31DEC2014 and the previous time period has date=31DEC2014 in cross section CONM=AGELLAN COMMERCIAL REIT.

However, it would be really helpful if you could provide me with an example of proc timeseries, to try your solution

thank you !!

Re: error proc panel :

sbxkoenk — Sat, 24 Sep 2022 13:23:21 GMT

Hello @Am_fr ,

Here's the code :

( Remark that PROC TIMESERIES is making the timestamp equally spaced, i.e. it fills gaps (with missings when using setmissing=missing) )

proc sort data=work.have 
     /* no out=  needed if input and output datasets are the same */ ;
 by cross_section timestamp;
run;

proc timeseries data=work.have out=work.dseries;
   by cross_section;
   id timestamp interval=day
                accumulate=median
                setmiss=missing
             /* start='01jan1998'd */
             /* end  ='31dec2000'd */ ;
   var item1-item8;
run;
/* end of program */

If it works now, you might want to use another accumulation statistic.

See here for your choices :

SAS/ETS 15.2 User's Guide
The TIMESERIES Procedure
ID Statement
Look for the ACCUMULATE=option !
https://go.documentation.sas.com/doc/en/pgmsascdc/9.4_3.5/etsug/etsug_timeseries_syntax08.htm

Cheers,

Koen