How to extract only numeric observations from a mixed variable list(num+char) in sas

[annypanny]

In a dataset there is a varible list where numeric as well as character values are there, how can I get only numeric observations from it in sas. example dataline are here:in only one variable list VAR1 there are values e.g:

variable name:

VAR

................

observations:

123_854

xyz

456_672

abc

789_458

frf

233_562

gth

....

qxd

345_672

I have to extract only that numeric values e.g(233_562 etc) from that whole dataline, How can I get that.

edited.  Previously it was solved

[PeterClemmensen]

@annypanny Welcome to the SAS Community 🙂

Are these eight separate variables or one variable with eight values in it?

[annypanny]

there is one variable and there are many values, I have just written eight values here. Thanks

[Kurt_Bremser]

Are these values stored in a single variable, or in separate variables within a given observation?

[annypanny]

there is one variable and there are many values, I have just written eight values here. Thanks

[PeterClemmensen]

Does that mean you want to get rid of the commas as well? If so 

data _null_;
    var1 = '123, xyz, 456, abc, 789, frf, 233, gth';
    var2 = compbl(compress(var1, '', 'ka'));
    put var2=;
run;

[annypanny]

what if VAR1 has more than thousand values not eight values, then what can I do? Thanks in advance

[ed_sas_member]

Hi @annypanny 

I am a bit confused:

are our input data like this:

data have;
    var1 = '123, xyz, 456, abc, 789, frf, 233, gth';
run;

or like this?

data have;
	input VAR1 $;
	datalines;
123
xyz
456
abc
789
frf
233
gth
;
run;

In the second case, you can use the following code to output only observation containing numbers:

data want;
	set have;
	if prxmatch('/\d+\s*$/',VAR1)>0 then output;
run;

Best,

[annypanny]

the second one is the case and I want to extract only character values.

[ed_sas_member · Accepted solution]

Hi @annypanny 

So you can do this:

data want;
	set have;
	if prxmatch('/\d+\s*$/',VAR1)>0 then delete;
run;
proc print;

Best,

[annypanny]

thanks, It is working now

[ed_sas_member]

Awesome!

You're welcome @annypanny 

[annypanny]

and how we can do the vice versa i.e., how to extract a only numeric character which can also contain special characters like "_". example 1234_4567 in the above question

[annypanny]

this act as a solution for me thanks

[PeterClemmensen]

The number of values does not matter. See this

data _null_;
    var1 = '123, xyz, 456, abc, 789, frf, 233, gth,
            123, xyz, 456, abc, 789, frf, 233, gth,
            123, xyz, 456, abc, 789, frf, 233, gth,
            123, xyz, 456, abc, 789, frf, 233, gth,
            123, xyz, 456, abc, 789, frf, 233, gth';
    var2 = compbl(compress(var1, '', 'ka'));
    put var2;
run;