Hi All
I create a dataset by setting many datasets together. Is it possible to find out which dataset contributed the observation?
eg
Data Health;
set one
two
threee
four;
run;
I need to mark where each obs came from
Thanks
Steve
Hi ... yes, this is certainly less obscure than the LOG2 stuff to give 1, 2, 3, 4, etc. and also one of those "duh" moments ...
data all;
set one (in=_1) two (in=_2) three(in=_3);
from = find(catt(of _:),'1');
run;
Hi ... if you have V9.2 or later, you can use the INDSNAME option on the SET statement ...
data one two three;
retain x 1;
run;
data all;
set one two three indsname=dsn;
dset = dsn;
run;
x dset
1 WORK.ONE
1 WORK.TWO
1 WORK.THREE
How about:
data health;
set one (in=a)
two (in=b)
three(in=c)
four(in=d);
if a then from=1;
if b then from=2;
if c then from=3;
if d then from=4;
run;
Hi Guys
Thanks for the replies. Really liking the INDSNAME but unforunately don;t have 9.2..yet..still on 9.1.3
Just working on the suggestion from Linlin
Looks like it's gonna be the long way round but good to see 9.2 has a solution
Thanks for sharing your views
Steve
Hi ... with that approach, you could also consider (assuming there are no variables in the data sets that start with an underscore) ...
data one two three;
retain x 1;
run;
data all;
length from $3;
set one (in=_1) two (in=_2) three(in=_3);
from1 = catt(of _:);
from2 =
run;
from x
100 1
010 1
001 1
If you want the 1, 2, 3, 4 etc. instead of 1's and 0's, you could try (though there might be some less obtuse way to do this) ...
data all;
set one (in=_1) two (in=_2) three(in=_3);
from = log2(input(reverse(catt(of _:)),binary.))+1;
run;
x from
1 1
1 2
1 3
I'm thinkin' is there a REGEX that would change ONE TWO THREE to ONE(in=in1) TWO(in=in2) THREE(in=in3) and there probably is but I don't know without research and the rest.
So I try the following. I did not expect it to work and of course I get disqualified for using MONOTONIC. And the numbers change each time you run it but that doesn't matter because they just need to be different names for each IN=option, not necessarily 1 2 3.
8 %put NOTE: DATA=&data;
NOTE: DATA=one two three
9 %let data2=%sysfunc(compbl(%superQ(data)))%str( );
10 %let data2=%sysfunc(transtrn(%superq(data2),%str( ),%nrstr((in=_in%sysfunc(monotonic())) )));
11 %put NOTE: DATA2=%superq(data2);
NOTE: DATA2=one(in=_in1) two(in=_in2) three(in=_in3)
12
13 data all;
14 set &data2;
15 length in $8;
16 in = cats(of _IN:);
17 name = scan("&data",index(in,'1'));
18 run;
Obs i in name
1 1 100 one
2 2 100 one
3 1 010 two
4 2 010 two
5 1 001 three
6 2 001 three
Hi ... yes, this is certainly less obscure than the LOG2 stuff to give 1, 2, 3, 4, etc. and also one of those "duh" moments ...
data all;
set one (in=_1) two (in=_2) three(in=_3);
from = find(catt(of _:),'1');
run;
Hi
I used the
find(catt(of _:), '1') statement and works perfectly.
Really interesting to break it down and see how the in= works and storage of the _vars in the PDV ie sequentially.
Apologies for not looking into all the other given replies.
Thanks all
Steve
If the input files can be a numbered list, then the following could be used:
data want;
set test1-test3 indsname=fname;
from=input(compress(fname,,'dk'),3.);
run;
If you want table name:
data one two three; set sashelp.class; output one two three; run; option mprint mlogic symbolgen; %let t_list=work.one work.two work.three; %macro tlist; %let i=1; %let name&i=%scan(&t_list,&i,%str( )); %do %while(&&name&i ne ); %let t_name&i=&&name&i(in=in&i) ; %let i=%eval(&i+1); %let name&i=%scan(&t_list,&i,%str( )); %end; data all ; set %do j=1 %to &i-1; &&t_name&j %end; ; length indsname $ 40; select; %do k=1 %to &i-1; when(in&k) indsname="&&name&k"; %end; otherwise; end; run; %mend list; %tlist
Ksharp
Registration is now open for SAS Innovate 2025 , our biggest and most exciting global event of the year! Join us in Orlando, FL, May 6-9.
Sign up by Dec. 31 to get the 2024 rate of just $495.
Register now!
Learn how use the CAT functions in SAS to join values from multiple variables into a single value.
Find more tutorials on the SAS Users YouTube channel.
Ready to level-up your skills? Choose your own adventure.