topic Re: Calculate descriptive statistics on all variables in SAS Programming

Calculate descriptive statistics on all variables

JUMMY — Thu, 04 Oct 2018 16:47:37 GMT

proc means data=merge_bcbs N MEAN STDDEV MEDIAN MAX MIN maxdec=4;
	var type srxtype_n antif fracture;
			title '';
run;

400  proc means data=merge_bcbs N MEAN STDDEV MEDIAN MAX MIN maxdec=4;
401      var type;
402              title '';
403  run;

NOTE: There were 100 observations read from the data set WORK.MERGE_BCBS.
NOTE: PROCEDURE MEANS used (Total process time):
      real time           0.03 seconds
      cpu time            0.03 seconds


404  proc means data=merge_bcbs N MEAN STDDEV MEDIAN MAX MIN maxdec=4;
405      var type srxtype_n antif fracture;
ERROR: Variable srxtype_n in list does not match type prescribed for this list.
ERROR: Variable antif in list does not match type prescribed for this list.
ERROR: Variable fracture in list does not match type prescribed for this list.
406              title '';
407  run;

NOTE: The SAS System stopped processing this step because of errors.
NOTE: PROCEDURE MEANS used (Total process time):
      real time           0.01 seconds
      cpu time            0.01 seconds

I am trying calculate descriptive statistics on all variables I created (inserted above in proc step). I keep getting an error message that the variable does not match the prescribed list. Most of my variables are character. How do I run proc means on character variables. IS there a way to convert it to numeric? What am I doing wrong?

Re: Calculate descriptive statistics on all variables

Reeza — Thu, 04 Oct 2018 16:38:14 GMT

You can't compute statistics on a character variable. SAS likely read a column in as character, when you want it as numeric.

Run a PROC CONTENTS and check your variable types.

@JUMMY wrote:

proc means data=merge_bcbs N MEAN STDDEV MEDIAN MAX MIN maxdec=4;
	var type srxtype_n antif fracture;
			title '';
run;

400  proc means data=merge_bcbs N MEAN STDDEV MEDIAN MAX MIN maxdec=4;
401      var type;
402              title '';
403  run;

NOTE: There were 100 observations read from the data set WORK.MERGE_BCBS.
NOTE: PROCEDURE MEANS used (Total process time):
      real time           0.03 seconds
      cpu time            0.03 seconds


404  proc means data=merge_bcbs N MEAN STDDEV MEDIAN MAX MIN maxdec=4;
405      var type srxtype_n antif fracture;
ERROR: Variable srxtype_n in list does not match type prescribed for this list.
ERROR: Variable antif in list does not match type prescribed for this list.
ERROR: Variable fracture in list does not match type prescribed for this list.
406              title '';
407  run;

NOTE: The SAS System stopped processing this step because of errors.
NOTE: PROCEDURE MEANS used (Total process time):
      real time           0.01 seconds
      cpu time            0.01 seconds

Re: Calculate descriptive statistics on all variables

PeterClemmensen — Thu, 04 Oct 2018 16:39:09 GMT

Please run

proc contents data=merge_bcbs;
run;

What types of variables are these?

Re: Calculate descriptive statistics on all variables

mkeintz — Thu, 04 Oct 2018 16:39:30 GMT

It appears that the unprocessed variables are of type character, not type numeric. Proc means is intended to generates stats only for numeric variables.

Re: Calculate descriptive statistics on all variables

Reeza — Thu, 04 Oct 2018 16:40:43 GMT

And if you exclude a VAR statement entirely, SAS will run it for all numeric variables by default.

Re: Calculate descriptive statistics on all variables

JUMMY — Thu, 04 Oct 2018 16:45:44 GMT

@PeterClemmensen. they are character variables. Is there a way to convert them to numeric and then do a proc means?

Re: Calculate descriptive statistics on all variables

JUMMY — Thu, 04 Oct 2018 16:46:21 GMT

@mkeintz, they are character variables. Is there a way to convert them to numeric and then do a proc means?

Re: Calculate descriptive statistics on all variables

mkeintz — Thu, 04 Oct 2018 17:04:09 GMT

If they are character variables, let's assume they contain nothing but blanks, contiguous numeric characters, and maybe one decimal point. One can deal with fancier character vars (say with currency symbols, commas, etc.) but the point here is to demonstrate the INPUT function, usually (but not exclusively) used to convert a character value into a numeric value:

data vneed / view=vneed;
  set have (rename=(srxtype_n=chrv1 antif=chrv2 fracture=chrv3));
  srxtype_n=input(chrv1,best32.);
  antif_n=input(chrv2,best32.);
  fracture_n=input(chrv3,best32.);
  put (_character_) (=) / (_numeric_) (=);
run;

proc means data=vneed;
  var ...... ;
run;

A couple of notes:

YOu can't convert a character variable to a numeric variable. I.e. a variable can only be one type for the duration of a data step. That's why I did renames of the character variables above. Then when the INPUT function was run, I could re-use the original variable names that you probably want to use in the proc means.
The input function needs 2 arguments, the character variable name, and the format to apply. I suspect that best32. format will serve your needs.
(edited additional note). Also, for efficiency's sake I made VNEED a data set "view", not a data set "file". Assuming you don't need the numeric conversion to be permanent, there is no need to write data set need to a disk only to re-read it once. A data set view avoids that process - instead it read from the original data set (HAVE), and transfers the content of data set need, one record at a time, to the proc means.

Re: Calculate descriptive statistics on all variables

JUMMY — Thu, 04 Oct 2018 17:12:13 GMT

@mkeintz, I am a new sas user. Pardon me. So what variables do I put under proc means?

Re: Calculate descriptive statistics on all variables

ballardw — Thu, 04 Oct 2018 17:12:14 GMT

@JUMMY wrote:
@mkeintz, they are character variables. Is there a way to convert them to numeric and then do a proc means?

Go back to when you read the data into SAS. Read the values as intended. If the source data was Excel convert to CSV first.

With CSV you can use Proc Import option guessingrows to read more rows before guessing the variable type, use a large value such as 32000.

Or write a data step or modify the data step code created by Proc Import, copy from the log and edit.

Re: Calculate descriptive statistics on all variables

mkeintz — Thu, 04 Oct 2018 17:21:47 GMT

That's a question easily answered by (1) reading one of @Reeza's comments, or (2) experimenting (which I believe all programmers do - so welcome to the club).