topic Re: Compress a numeric "." in SAS Programming

Compress a numeric "."

Mathis1 — Fri, 04 Sep 2020 10:56:43 GMT

Hi,

I have a numeric variable taking the values 0,1 and "." . When using the function compress or cats for this variable, the "." is tranformed into a 0. Is there any way to avoid this aside from transforming the numeric variable into a character variable taking the value "0" , "1" and ".".

Thank you 😉

Re: Compress a numeric "."

Tom — Fri, 04 Sep 2020 11:12:20 GMT

The COMPRESS() function operates on character strings. Why would you ever use it on a numeric value?

Re: Compress a numeric "."

Kurt_Bremser — Fri, 04 Sep 2020 11:15:50 GMT

CATS and COMPRESS are character functions, which do not work well with numeric variables.

A missing value in SAS is, per default, represented as a dot. You can change how SAS displays a missing value with the MISSING= system option. If you get a 0 instead of the dot when using the implicit conversion from numeric to character, then most likely your MISSING option is set to a zero.

For detailed help, post your code and log, and some example data. Please post your example data as a data step with datalines, so it is easy for us to recreate your dataset in our environment.

Use this button:

for posting logs (and other textual data that must be kept as-is), and the "little running man" right next to it for SAS code.

Re: Compress a numeric "."

RichardDeVen — Fri, 04 Sep 2020 12:01:31 GMT

How a missing value is rendered in output is controlled with the system option MISSING.

You can examine the current setting of the option by looking in the LOG

Example:

proc options option=MISSING;
run;

Log

133  proc options option=missing;
134  run;

    SAS (r) Proprietary Software Release 9.4  TS1M6

 MISSING=.         Specifies the character to print for missing numeric values.
NOTE: PROCEDURE OPTIONS used (Total process time):

The option allows you to specify the single character to use.

Example:

data have;
  input x @@; datalines;
0 1 .
;

options missing = . nosource;

proc options option=missing;
run;

data _null_;
  set have;
  result = cats (x);
  put 'NOTE:' x= result=;
run;

options missing = 0;   * Show missing values as 0 instead of default .;

data _null_;
  set have;
  result = cats (x);
  put 'NOTE:' x= result=;
run;

options missing = !;  * Show missing values as exclamation point (!);

data _null_;
  set have;
  result = cats (x);
  put 'NOTE:' x= result=;
run;

options missing = . source;

Log (the X= value is the missing value rendering, and the RESULT= value is the rendered missing stored in a variable)

    SAS (r) Proprietary Software Release 9.4  TS1M6

 MISSING=.         Specifies the character to print for missing numeric values.
NOTE: PROCEDURE OPTIONS used (Total process time):
      real time           0.00 seconds
      cpu time            0.00 seconds



NOTE:x=0 result=0
NOTE:x=1 result=1
NOTE:x=. result=.        <---------------
NOTE: There were 3 observations read from the data set WORK.HAVE.
NOTE: DATA statement used (Total process time):
      real time           0.00 seconds
      cpu time            0.00 seconds



NOTE:x=0 result=0
NOTE:x=1 result=1
NOTE:x=0 result=0         <---------------
NOTE: There were 3 observations read from the data set WORK.HAVE.
NOTE: DATA statement used (Total process time):
      real time           0.00 seconds
      cpu time            0.00 seconds



NOTE:x=0 result=0
NOTE:x=1 result=1
NOTE:x=! result=!         <---------------
NOTE: There were 3 observations read from the data set WORK.HAVE.
NOTE: DATA statement used (Total process time):
      real time           0.00 seconds
      cpu time            0.00 seconds

Re: Compress a numeric "."

ballardw — Fri, 04 Sep 2020 15:47:15 GMT

Large economy sized hint:

Provide some example data and how you want the output to appear.

Since you mention CATS function then that seems to imply that you have multiple variables.

Here's a start of data set with 3 variables and all combinations of ., 0, and 1.

data have;
   input v1 v2 v3;
datalines;
. . .                                                           
. . 0                                                           
. . 1                                                           
. 0 .                                                           
. 0 0                                                           
. 0 1                                                           
. 1 .                                                           
. 1 0                                                           
. 1 1                                                           
0 . .                                                           
0 . 0                                                           
0 . 1                                                           
0 0 .                                                           
0 0 0                                                           
0 0 1                                                           
0 1 .                                                           
0 1 0                                                           
0 1 1                                                           
1 . .                                                           
1 . 0                                                           
1 . 1                                                           
1 0 .                                                           
1 0 0                                                           
1 0 1                                                           
1 1 .                                                           
1 1 0                                                           
1 1 1  
;

So now you just have to show what the result should be.

Re: Compress a numeric "."

Mathis1 — Mon, 07 Sep 2020 13:19:10 GMT

Hi,
If I take your example ballardw, I would like to create a new variable V4 for which the first line would be :
V1 V2 V3 V4
. . . ._._.

Where "_" is the delimiter

Re: Compress a numeric "."

Tom — Mon, 07 Sep 2020 13:35:00 GMT

The CATX() function should so that. Make sure the missing option is set to '.' and not something else, like '0' or ' '.

%let missing=%sysfunc(getoption(missing));
options missing='.';
data have;
 do v1=.,0,1; do v2=.,0,1; do v3=.,0,1;
   v4 = catx('_',of v1-v3);
   output;
 end;end;end;
run;
options missing="&missing";

proc print;
run;

Obs    v1    v2    v3     v4

  1     .     .     .    ._._.
  2     .     .     0    ._._0
  3     .     .     1    ._._1
  4     .     0     .    ._0_.
  5     .     0     0    ._0_0
  6     .     0     1    ._0_1
  7     .     1     .    ._1_.
  8     .     1     0    ._1_0
  9     .     1     1    ._1_1
 10     0     .     .    0_._.
 11     0     .     0    0_._0
 12     0     .     1    0_._1
 13     0     0     .    0_0_.
 14     0     0     0    0_0_0
 15     0     0     1    0_0_1
 16     0     1     .    0_1_.
 17     0     1     0    0_1_0
 18     0     1     1    0_1_1
 19     1     .     .    1_._.
 20     1     .     0    1_._0
 21     1     .     1    1_._1
 22     1     0     .    1_0_.
 23     1     0     0    1_0_0
 24     1     0     1    1_0_1
 25     1     1     .    1_1_.
 26     1     1     0    1_1_0
 27     1     1     1    1_1_1