Proc Means and Proc Univariate Differences

[turcay]

Hello,

I wonder, do proc means procedure's and proc univariate procedure's functions calculate the same values. Firstly i wrote the procedure with using proc means now i want to try it with proc univariate but i got error i think their syntax is different. How can i write proc means code  as proc univariate

proc means data=WORK.sample STACKODS
FW=12
     PRINTALLTYPES
     CHARTYPE
     QMETHOD=OS
     VARDEF=DF 
           MEAN 
           STD 
           MODE    
           P10 
           P90 ;
var Q:;
ods output summary=stacked;
run;

proc univariate data=WORK.sample STACKODS
FW=12
     PRINTALLTYPES
     CHARTYPE
     QMETHOD=OS
     VARDEF=DF 
           MEAN 
           STD 
           MODE    
           P10 
           P90 ;
var Q:;
ods output summary=stacked;
run;

Thank you.

[ballardw]

Please look in the online help as there are significant differences in syntax. As a hint see what you get with

proc univariate data=WORK.sample ;
var Q:;

run;

Printed appearance of decimal values may appear different due to different default widths of values displayed.

[turcay]

Yes of course tried the statement that you wrote but ı try to get same output like below.

Thank you.

[Reeza]

I don't think you easily can. Proc Univariate does not support the same output format in general. What do you want from proc univariate that you can't get in proc means?

[turcay]

Okay then. Some of my friends said to me,univariate can generate different values but he is not sure, as you said there is no differences. I wanted to be sure. Thanks. Well,  how can we generate same output with univariate ? 

Thank you.

[Reeza]

You can use the OUTPUT statement to generate the dataset and then you'd have to transpose it using proc transpose.

Or you can capture several of the ODS tables, merge them and/or transpose.

The documentation is fairly clear on how to get output from PROC UNIVARIATE, unlike proc means you do need to explicitly list variable names as well.

Proc univariate will generate same values for the same statistics, but it will also generate specific percentiles points which is when I generally use it over proc means.

https://support.sas.com/documentation/cdl/en/procstat/63104/HTML/default/viewer.htm#procstat_univari...

[Ksharp]

You can use the following code to get the same layout as proc means,but you can't customize it .

proc univariate data=sashelp.class outtable=want;
var weight;
run;

[turcay]

Thank you , i'm so close to do it. Actually, at the first code i made it but i didn't bring the columns which i want it. Second code i got the columns but just for one Q. Have you got an any idea ?.

bdata sample;

length Q1 8 Q2 8 Q3 8 Q4 8 Q5 8 Q6 8;

infile datalines  missover dlm=',';

input Q1 Q2 Q3 Q4 Q5 Q6;

datalines;

80,90,70,90,80,70

90,95,99,40,50,90

0,90,99,89,87,89

77,88,0,45,78,89

0,0,0,58,90,89

-10,-10,-30,89,90,79

-20,-45,-80,89,90,50

-49,-67,-80,70,89,80

;

proc univariate data=work.sample outtable=sample;

   var Q:;

run;

proc univariate data=work.sample2;

   var _VAR_ _MEAN_  _STD_ _P10_ _P90_ _MODE_;

run;

proc univariate data=work.sample;

   var Q:;

   output out=univariateTable

mean=Mean

Mode=Mode

P10=P10

P90=P90

STDDEV=STDDEV;

run;

[Ksharp]

The table (want) generated by OUTTABLE= have already contain these statistic estimator.

You just need KEEP it .

data sample;
length Q1 8 Q2 8 Q3 8 Q4 8 Q5 8 Q6 8;
infile datalines  missover dlm=',';
input Q1 Q2 Q3 Q4 Q5 Q6;
datalines;
80,90,70,90,80,70
90,95,99,40,50,90
0,90,99,89,87,89
77,88,0,45,78,89
0,0,0,58,90,89
-10,-10,-30,89,90,79
-20,-45,-80,89,90,50
-49,-67,-80,70,89,80
;

proc univariate data=work.sample outtable=want(keep=_var_ _mean_ _std_ _mode_ _p10_ _p90_);
   var Q:;
run;

[turcay]

Yes I made it, is it possible to change columns name ? I tried but did not succeed.

Thank you.

proc univariate data=work.sample outtable=want(keep=_var_ _mean_ _std_ _mode_ _p10_ _p90_)
;
label _VAR_=Variable;
   var Q:;
run;

[Tom]

Yes. if you wanted to rename _VAR_ to VARIABLE you would  include it in the list of old=new names in the RENAME option on the output dataset name.

PROC UNIVARIATE data=sashelp.class
  OUTTABLE=want
   (
     RENAME=(_VAR_ = VARIABLE
            )
   )
 ;
RUN;

[turcay]

Yes it works tahnk you. But this time i couldn't  use drop or keep statement ?

PROC UNIVARIATE data=WORK.sample
  OUTTABLE=want (keep=_var_)
  (RENAME=(_VAR_ = VARIABLE));
  var Q:;
RUN;

Thank you.

[Tom · Accepted solution]

You have an extra close parenthesis. Remove the right parenthesis after your list of variables to keep. Dataset options are enclosed with () following the dataset name.   They are space delimited and in the form of option=value.  Some like the RENAME= option require their own set of () since the values contain equal signs. Formatting the code can make it easier to insure that they are properly nested.

mydsn
(keep= x y x 
 rename=
    (x=a
     y=b
    ) 
 obs=10 
 firstobs=3 
)

[turcay]

Yes. That's it. Thank you 🙂

PROC UNIVARIATE data=WORK.sample OUTTABLE=want
 (keep=_var_ _mean_ _std_ _mode_ RENAME=(
_VAR_=VARIABLE
_MEAN_=MEAN
_STD_=STDDEV
_MODE_=MODE
));
  var Q:;
RUN;