Subset using proc contents output

[Lopa2016]

Hi,

Is it possible to subset a data set based on the proc contents output. For e.g. I want to exclude all those variables which are a part of another data set .

Thanks in advance.

[Kurt_Bremser]

Retrieve meta-information about datasets from dictionary.tables (SQL) or sashelp.vtable (data step). This will then help you in creating macro variables for use in drop or keep statements.

[PeterClemmensen · Accepted solution]

If you insist on using PROC CONTENTS output then do like below. 

Otherwise, i agree with @Kurt_Bremser, get the information from dictionary tables or sashelp.vtable instead.

proc contents noprint data=sashelp.heart out=VarsToDrop(keep=name);
run;

proc sql noprint;
	select name into :vars separated by ' '
	from VarsToDrop;
quit;

%put &vars.;

data ClassExludingVars;
	set sashelp.class;
	drop &vars.;
run;

[mkeintz]

Let's say you want all the variables in B that are not also in A:

data bvars_only;
  retain _sentinel1;
  if 0 then set a;
  retain _sentinel2;
  set b;
  drop _sentinel1--_sentinel2;
run;

This relies on how SAS constructs the program data vector. This program orders variables in the PDV from left to right as

1. _sentinel1
2. all the A vars
3. _sentinel2
4. all the B vars not already in item2 above.

The drop statement, using a double-dash connector tells SAS to drop all the vars in item1, item2, and item3.

[Astounding]

Let's say you want to drop variables in B which also appear in A.  You could start with:

proc contents data=A noprint out=contents_A (keep=name);

run;

proc contents data=B noprint out=contents_B (keep=name);

run;

Now the issues are (1) finding variable names that appear in both, despite the fact that (2) capitalization of the names can be different in both data sets.  So:

data contents_A;

set contents_A;

name = upcase(name);

run;

proc sort data=contents_A;

by name;

run;

data contents_B;

set contents_B;

name = upcase(name);

run;

proc sort data=contents_B;

by name;

run;

Having a list of variable names from both sources (capitalized and sorted), the next step is to find those names appear in both:

data matching;

merge contents_A (in=ina) contents_B (in=inb);

by name;

if ina and inb;

run;

Having found the matches, extract the names into a macro variable:

proc sql;

select name into : matching_vars separated by ' ' from matching;

quit;

Finally, drop the variables:

data want;

set a (drop=&matching_vars);

run;

Unfortunately, it takes a long series of steps (at least the way that I'm picturing the problem requires many steps).

[s_lassen]

It is easier to use dictionary.columns to get the common variables:

proc sql noprint;
  select name into :dropvars separated by ' '
  from dictionary.columns
  where libname='SASHELP' and memname='CARS'
   and name in(select name from dictionary.columns where libname='SASHELP' and memname='CLASS');
  ;
quit;

data class_without_cars;
  set sashelp.class(drop=&dropvars);
run;

This will drop the variable WEIGHT, which is common to the two datasets.