Proc summary

[hhchenfx]

Good morning,

I have a data with a Target variable and a number of condition variable ( a b c d e f).

All variable having value 1 or 0.

I want to analyze how many observation of Target=1 (and Target=0) for each combination of 2 condition value (a=1, b=1; a=1, b=0...).

My problem is I dont know how to tell SAS to do ab ac ad.... using the proc summary.

The output should look like the one below. However to get it, I have to go a long way and inefficient.

Thank you for your help.

HHC

Obs    a    b    c    Target    _TYPE_    _FREQ_

                                                                                               1    .    0    0       0       0111        2

                                                                                               2    .    0    0       1       0111        1

                                                                                               3    .    0    1       1       0111        1

                                                                                               4    .    1    1       0       0111        2

                                                                                               5    .    1    1       1       0111        2

                                                                                               6    0    .    0       0       1011        1

                                                                                               7    0    .    1       1       1011        2

                                                                                               8    1    .    0       0       1011        1

                                                                                               9    1    .    0       1       1011        1

                                                                                              10    1    .    1       0       1011        2

                                                                                              11    1    .    1       1       1011        1

                                                                                              12    0    0    .       0       1101        1

                                                                                              13    0    0    .       1       1101        1

                                                                                              14    0    1    .       1       1101        1

                                                                                              15    1    0    .       0       1101        1

                                                                                              16    1    0    .       1       1101        1

                                                                                              17    1    1    .       0       1101        2

                                                                                              18    1    1    .       1       1101        1

data
have;

     input
Target a b c d e f ;

     datalines;

0 1 1 1 1 1 1

  1 0 0 1 1 0 0

1 1 0 0 0 1 1

0 1 1 1 1 0 0

0 0 0 0 1 0 0

1 0 1 1 0 1 0

0 1 0 0 0 0 0

1 1 1 1 1 1 1

  ;;;;

    run;

[Reeza]

Same answer as before, look up the options in WAYS and TYPES so you understand what it's doing.

Then develop your filter criteria for the output data set either using a where statement and the _type_ variable or using TYPES statement.

data have;

input Target a b c d e f ;

datalines;

0 1 1 1 1 1 1

1 0 0 1 1 0 0

1 1 0 0 0 1 1

0 1 1 1 1 0 0

0 0 0 0 1 0 0

1 0 1 1 0 1 0

0 1 0 0 0 0 0

1 1 1 1 1 1 1

;

data have2;

    set have;

    count=1;

run;

proc means data=have2 noprint;

class a b c d e f target;

var count;

ways;

output out=test1 (where=(target ne .)) sum(target)=sum_target;

run;

[hhchenfx]

Thanks, Reeza.

It is kind of similar to my approach below where we tell SAS not to include target=.

What make me uncomfortable is we treat target and all condition variable the same at the beginning and for a large data, I afraid it will create a lot of work.

proc summary data=have chartype;

class a b c d e f target;

ways 3;

output  out=sumtest;

run;

data sumtest; set sumtest;

if target=. then delete; run;

[Reeza]

hhchenfx wrote:

What make me uncomfortable is we treat target and all condition variable the same at the beginning and for a large data, I afraid it will create a lot of work.

Can you explain the above further, I don't understand.

[data_null__ · Accepted solution]

What if you combine CLASS TYPES and BY TARGET; The only difference the value of _TYPE_, but it is 1 for all target anyway.

data have;
   input target a b c d e f ;
   datalines;
0 1 1 1 1 1 1
1 0 0 1 1 0 0
1 1 0 0 0 1 1
0 1 1 1 1 0 0
0 0 0 0 1 0 0
1 0 1 1 0 1 0
0 1 0 0 0 0 0
1 1 1 1 1 1 1
  ;;;;
    run;
proc sort data=have;
   by target;
   run;
proc summary data=have chartype;
   by target;
   class a b c d e f;
   ways 2;
   output out=bytar1;
   run;
proc sort;
   by _type_ target;
   run;
proc summary data=have chartype;
   class target a b c d e f;
   ways 3;
   output out=tar1(where=(first(_type_) eq '1'));
   run;
proc compare base=bytar1(drop=_type_) compre=tar1 listequalvars;
   run;

[hhchenfx]

Thank you, Data_null_.

I think it works nicely.

HHC