topic SAS Vertical Merge in SAS Procedures

SAS Vertical Merge

arjunaj91 — Sun, 22 Nov 2020 03:47:17 GMT

I have three tables and I want to vertically merge those tables

Table A

Name Money

X 2000

Y 4000

Table B

Name Money

X 600

Z 400

Table C

Name Money

Y 8000

Z 600

I want to combine these tables into single table like this using proc sql so that I can save the table:

Name Money_1 Money_2 Money_3

X 2000 600 0

Y 4000 0 8000

Z 0 400 600

Like this I have 50 tables and I want to merge them vertically using proc sql

One solution is joining two tables at once using full join but in select statement choose the Name variable of table with the most rows. That works for few tables but i have over 50 tables and I need a feasible solution to this problem

Re: SAS Vertical Merge

Kurt_Bremser — Sun, 22 Nov 2020 04:58:34 GMT

There is no "vertical merge". You can either stack datasets vertically, or join/merge them horizontally.

What you want can be most simply achieved by first stacking, with an interleave, and then transposing:

data long;
length dname $41 dsname $32;
set
  table_a
  table_b
  table_c
  indsname=dname
;
by name;
dsname = scan(dname,2,"."); /* removes library name */
run;

proc transpose
  data=long
  out=want
;
by name;
var money;
id dsname;
run;

With additional logic in the data step, you can create variable names for the transpose to your liking.

The only thing that should be automated is the list of dataset names for the SET statement. You can use a colon wildcard there (lib.table:), or retrieve the list from DICTIONARY.TABLES in SQL into a macro variable.

Re: SAS Vertical Merge

Tom — Sun, 22 Nov 2020 05:15:14 GMT

I would just leave it in a tall structure. You can then produce that report using a reporting procedure like PROC REPORT, TABLULATE or FREQ.

data tall;
  length indsname dsname $41 ;
  set A B C indsname=indsname;
  by name;
  dsname=indsname;
run;

Obs    dsname    Name    Money

 1     WORK.A     X       2000
 2     WORK.B     X        600
 3     WORK.A     Y       4000
 4     WORK.C     Y       8000
 5     WORK.B     Z        400
 6     WORK.C     Z        600

proc freq ;
 tables name*dsname;
 weight money ;
run;