I am trying to crash course my way through a hash solution with little luck.  I have 5 different account numbers and I am looking to join the first(acctnbr) to any of the following 4.  Then I will sort by acctnbr, opendate desc to find the oldest original account number associated with any real account.  I can use name or ss# to confirm the accuracy.  The first dataset below has 8 mill records, the second has 16 million, and these are representing 3 months, later I will be using 12 months.  Is HASH a viable solution for this many records?  If not I have a another sql join below but I'm not sure that will get the join I am looking for.  If you can take a look at the attempt at the sql join I would appreciate that as well.      data consumer_agg; set consumer; keep acctnbr origacct opendate; run; data fdr_agg; set fdr; keep cracct xref1 xref2 xrefacct opendate; run; data orig_opendate;     length acctnbr $20 origacct $20 opendate 8;     if _n_ = 1 then do;  declare hash e(dataset: 'work.fdr_agg');  e.definekey ('acctnbr');  e.definedata('origacct','opendate','cracct','xref1','xref2','xrefacct');  e.definedone();  end;  set consumer_agg;  drop rc;  rc=e.find(); run;     /*********************/ sql solution:   data test1; set consumer_agg(obs=100); run; data test2; set fdr_agg(obs=100); run; proc sql;                     create table cartesian_join as         select distinct a.acctnbr,a.origacct,a.opendate as cons_date,b.*                         from test1 a,           test2 b order by opendate desc,cons_date desc; quit; ... View more

I am trying to join a years worth of tables together while adding a field 'Reprting_Month'.  What I have isn't working, I'm hoping it'll make enough sense to get the point accross.     What I have; consumer201611 field1 field1   consumer201610 field1 field1   What I want:   Consumer_Agg 201611 field1 field2 201610 field1 field2.   What I've tried so far is somethig like this:   data consumer_agg; %macro dates; set %do date = 201511 %to 201611; fdr.consumer&dates; %end; %mend; end;   the code runs with no errors but has no output.  I know I've done this in the past but can't find my old code.   ... View more

This is my first post here, usually over in Base Programming, I'm hoping a few of the moderators from there see this and answer as I respect their opinions greatly.    I have several years SAS programming experience but am new to statistical analysis / modeling.  I am finding the business people that I need to get the support of don't understand anything but decision trees and regression, really they don't understand regression but they've been beaten into submission and will accept it.  So... from that perspective, as a student I am learning a lot of new techniques, some that are still being developed today.  I wonder how to communicate to people that this is valuable information.  People that I am looking to for mentorship have given up communicating and now dumb down their work to their audience.    I would love to hear perspectives from people that are in this situation, have submitted to it, are working through it, or have conquered it.  Cheers, Mark ... View more

I am using the below code, the intent is to verify that the sum/difference of three columns equals another column.  I am not getting the intended output.    I also tried the commented line but got the same output:   data rules; set other_data; format delta value $20.; if not missing(d_net_co) then do; /*if D_NET_CO ne (D_GCO + D_BKCO - D_RECOVERIES) then*/ if D_NET_CO - (D_GCO + D_BKCO - D_RECOVERIES) ne 0 then edit_67 = 1;end; if edit_67 = 1; Value = (D_GCO + D_BKCO - D_RECOVERIES); Delta = value - d_net_co; keep rssd_id segment_id d_net_co d_gco d_bkco d_recoveries value delta; run;   Please find the attachment that shows the output.    Can someone help me understand why the delta column containing zero's are included with the output?   ... View more