topic Re: how can I extract name of correlated variables? in Statistical Procedures

how can I extract name of correlated variables?

marzieh — Sun, 25 Dec 2011 07:55:51 GMT

I have a data set like

A B C D E

1 12 3.2 ....

1.5 10 4

1.3 11 5.6

I want to extract name of correlated variables from it

for example I want a result like this:

A B

C D

which means A&B are correlated, C&D are correlated and E in independent

I used proc corr to have correlation matrix but I don't know how can I extract name of correlatd variables from it.

Can anybody help me in this issue?

Thanks a lot

Re: how can I extract name of correlated variables?

data_null__ — Sun, 25 Dec 2011 12:44:10 GMT

This works but It seems like I remember an easier way, oh well. :smileyplain:

ods output  PearsonCorr= PearsonCorr;
proc corr data=sashelp.heart;
   run;
ods output close;
 
proc transpose out=probs;
   by variable notsorted;
   var PAgeCHDdiag--PCholesterol;
   run;
data probs(keep=Variable With);
   set probs;
   by variable notsorted;
   if first.variable then do;
      i + 1;
      j = 0;
      end;
   j + 1;
   if i ge j then delete; *Upper;
   if col1 gt .05 then delete; *Cutoff;
   length With $32;
   with = substr(_name_,2);
   run;
options byline=1;
proc print;
   by variable notsorted;
   id variable;
   run;

Re: how can I extract name of correlated variables?

Linlin — Sun, 25 Dec 2011 16:13:28 GMT

data have;

infile cards missover;

input A B C D E;

cards;

1 12 3.2 . .

1.5 10 4 . 8

1.3 11 5.6 7 9

;

run;

data ab (rename=(a=var1 b=var2));

set have(keep=a b);

i=_n_;

group=1;

data cd (rename=(c=var1 d=var2));

set have(keep=c d);

i=_n_;

group=2;

data e (rename=(e=var1));

set have(keep=e);

i=_n_;

group=3;

data want (keep=var:);

set ab cd e;

by i group;

run;

proc print;

run;

Linlin

Re: how can I extract name of correlated variables?

Ksharp — Mon, 26 Dec 2011 07:15:33 GMT

You need variable cluster method.

proc varclus

Ksharp