09-15-2015 07:36 AM
I'm disappointed with proc tabulate : I use it with a rather small table, dynamic number of colums selected.
As soon as 6 columns are used I get an error :
Limit of 2147483647 potential interactions within a crossing exceeded for the table statement
How does he get from 16704 observations to such a limit ????
I tried using classsdata 'cause I thought it would make it easier on the tabulate, but it doens't help.
proc tabulate data=work.DATA_EXTRACT_report_sum missing classdata=work.DATA_EXTRACT_classlevel;
class barema_PC_6_code contract_zetel_code contract_zetel_oms beheerder_nr beheerder_naam maand barema_gebruik;
var firma_aant firma_aant_barema fdcp_aant fdcp_aant_barema;
table ( (barema_PC_6_code * (contract_zetel_code * (contract_zetel_oms * (beheerder_nr * (beheerder_naam ) ) ) all='Total') ) ) ALL
, ( maand * barema_gebruik )
* ( firma_aant*SUM=' ' firma_aant_barema*SUM=' ' fdcp_aant*SUM=' ' fdcp_aant_barema*SUM=' ' );
ERROR: Limit of 2147483647 potential interactions within a crossing exceeded for the table statement at line 2.
NOTE: The SAS System stopped processing this step because of errors.
NOTE: There were 16754 observations read from the data set WORK.DATA_EXTRACT_REPORT_SUM.
NOTE: There were 16754 observations read from the data set WORK.DATA_EXTRACT_CLASSLEVEL.
NOTE: PROCEDURE TABULATE used (Total process time):
real time 0.20 seconds
cpu time 0.20 seconds
09-15-2015 10:08 AM
Have you taken a look at Problem Note 21125: PROC TABULATE incorrectly issues limit of potential interactions error with ODS ... ?
It may also be that the cardinality of your class variables causes the overrun.
09-15-2015 10:17 AM
Think of how many discrete values your class variables have.
If you have 7 variables with 20 values each, you end up with 20**7 combinations. 30 discrete values, and you are at 21 million combinations. Since TABULATE does all that in memory, there's a limit imposed on the size of the structures required.
09-15-2015 10:41 AM