topic Re: Summary of statistics for ALL Combination from a set of variables in SAS Programming

Summary of statistics for ALL Combination from a set of variables

hhchenfx — Wed, 29 Jan 2014 02:29:21 GMT

Hi Everyone,

I have a dataset with a variable name target (taking value 0 or 1) and n independent variables.
I want to create a summary file that reports number of target=0 and number of target =1 for each combination of independent value.

data have;

input target a1 a2 a3 a4 a5 a6;

datalines;

0 1 9 1 0 8 1

1 1 0 1 1 5 0

1 1 9 2 3 1 1

1 2 3 0 2 0 6

0 2 9 1 7 0 0

0 3 3 0 9 0 3

1 2 1 1 2 1 2

0 1 2 0 3 0 4

;run;

Basically the summary will include All possible combination (from 1 factor to 6 factor from the 6 independent variables) such as:

If a1=1          how many observations have target=0 and how many have target=1.
If a1=1 and a2=9,       how many observations ....
If a1=1 and a2=9 and a3=1      how many observations ....
...
If a1=1 and a2=9 and a3=1 and a4=0 and a5=8 and a6=1      ....

If a1=2 how many observations ...
...
...

I don't know how to do it so any help is very much appreciated.

I am thinking about creating a condition file that summary distinct values for each independent variable and then run some kind of Do Loop to create the file I want.
The code for distinct value is as below just in case you might need.

Thank you for your help as always.

HHC

proc summary data=have missing chartype;
   class a:;
   ways 1;
   output out=distinct(drop=_type_ _freq_) / levels;
   run;
proc print;
   run;

proc sort data=distinct;
   by _level_;
   run;
data condition;
   update distinct(obs=0) distinct;
   by _level_;
   run;

Re: Summary of statistics for ALL Combination from a set of variables

data_null__ — Wed, 29 Jan 2014 03:19:23 GMT

Is it this?

data have;
  input target a1-a6;
  datalines; 
0 1 9 1 0 8 1
1 1 0 1 1 5 0
1 1 9 2 3 1 1
1 2 3 0 2 0 6
0 2 9 1 7 0 0
0 3 3 0 9 0 3
1 2 1 1 2 1 2
0 1 2 0 3 0 4
;;;;
run; 
proc sort; 
   by target;
   run; 
proc print; 
   run; 
proc summary data=have chartype missing; 
   by target;
   class a:;
   output out=combo;
   run; 
proc print; 
   run;

Re: Summary of statistics for ALL Combination from a set of variables

hhchenfx — Wed, 29 Jan 2014 14:07:18 GMT

Yes, it is, Data_null.

My intention is like that:

Get a subset of data by using the IF section, then work on that subset, export the result and go back to IF and work on another condition.

So your code is enable me to (1) got the Frequency and (2) got the list of all combination so that I can do the iteration for deeper analysis with each subsample.

Thank you.

HHC

Re: Summary of statistics for ALL Combination from a set of variables

hhchenfx — Wed, 29 Jan 2014 16:24:10 GMT

Hi Data_null,

Is there any quick change to your code if I want to change from "AND" to "OR" in my argument below?

Thank you,

HHC

If a1=1          how many observations have target=0 and how many have target=1.
If a1=1 OR a2=9,       how many observations ....
If a1=1 OR a2=9 OR a3=1      how many observations ....
...
If a1=1 OR a2=9 OR a3=1 OR a4=0 OR a5=8 OR a6=1      ....

If a1=2 how many observations ...