*_________________.
* steps in splitting the data model validation.
* Splitting the data into two parts.
*1) create a file with only the independent variables.  
* 2) standard the indep.
* 3) Run commands below.
*3a) Save data file - you will need to merge with original data.
* 4)  do vol.
* If you want a 75  - 25 split, then do the following.
*  duplex2 then vol.

set mxloop = 99999.
matrix.


get x/ variables = all /file = * /missing = omit /names = nam1.
 
save x /outfile = 'c:\temp1.sav' /names=nam1.
 
compute nr = nrow(x).
compute nc = ncol(x).
 
compute j = make(nr,1,1).
 
compute mean = t(j)*x/nr.
 
print mean.
 
compute diff = x-j*mean.
 
compute bigs = t(diff)*diff.
print bigs.
 
compute s1   =mdiag(diag(bigs)&**-.5).
print s1.
 
 
 
compute z =x*s1.
 
compute zpz =t(z)*z.
print zpz.
 
compute t1 = chol(zpz).
 
print t1.
 
compute diff1 = zpz -t(t1)*t1.
 
print diff1.
 
compute w = z*inv(t1).
 
compute di = t(w)*w.
print di.
 
*compute distances.
compute d1 = make(nc,1,0).
compute d2 = make(nc,1,0).
compute bign = nr*(nr-1)/2.
compute dist = make(bign,3,0).
 
compute iii=0.
 
loop ii = 1 to nr.
loop jj = ii+1 to nr.
compute iii = iii+1.
loop kk = 1 to nc.
compute d1(kk,1) = w(ii,kk).
compute d2(kk,1) = w(jj,kk).
end loop.
compute dist(iii,1) = ii.
compute dist(iii,2) = jj.
compute   dist(iii,3) = t(d1-d2)*(d1-d2).
end loop.
end loop.
 
save w/outfile = 'c:\w.sav'.
save dist/outfile = 'c:\dist.sav'.
 
end matrix.
GET
 FILE='C:\dist.sav'.
SORT CASES BY
 col3 (D) .
 
SAVE OUTFILE='C:\dist.sav'
 /COMPRESSED.
 
matrix.
 
get dist/variables = all /file = 'C:\dist.sav'.
get w /variables = all/ file = 'C:\w.sav'.
 
compute nr = nrow(w).
compute nc = ncol(w).
compute d1 = make(nc,1,0).
compute d2 = make(nc,1,0).
 
compute mean_x = make(nc,1,0).
compute mean_y = make(nc,1,0).
compute zero1 = make(nc,1,0).
compute j1 = make(nr,1,1).
 
compute in1 = make (nr,3,0).
loop ii = 1 to nr.
compute in1(ii,1) = ii.
end loop.
 
compute in1(dist(1,1),2) = 1.
compute in1(dist(1,2),2) = 1.
compute in1(dist(2,1),3) = 2.
compute in1(dist(2,2),3) =2.
 
* print in1.
 
 
compute mean_x = zero1.
compute ind1 = 0.
 
 
loop ii = 1 to nr.
 
do if (in1(ii,2) = 1).
compute ind1 = ind1 +1.
loop jj = 1 to nc.
*print ii.
*print jj.
*print ind1.
*print in1(ii,1).
*print w(in1(ii,1),jj).
 
compute mean_x(jj,1) = mean_x(jj,1)+w(in1(ii,1),jj).
end loop.
end if.
 
end loop .
compute mean_x = mean_x/ind1.
print mean_x.
 
 
 
compute ii = 0.
compute jj = 0.
compute mean_y = zero1.
compute ind1 = 0.
loop ii = 1 to nr.
do if (in1(ii,3) = 2).
compute ind1 = ind1 +1.
loop jj = 1 to nc.
*print ii.
*print jj.
*print ind1.
*print in1(ii,1).
*print w(in1(ii,1),jj).
 
compute mean_y(jj,1) = mean_y(jj,1)+w(in1(ii,1),jj).
end loop.
end if.
 
end loop .
compute mean_y = mean_y/ind1.
print mean_y.
 
* find distance.
 
compute dist1 = make(nr,2,-1).
 
loop ii = 1 to nr.
 
loop kk = 1 to nc.
compute d1(kk,1) = w(ii,kk).
compute d2(kk,1) = mean_x(kk,1).
end loop.
compute dist1(ii,1) = ii.
 
compute   dist1(ii,2) = t(d1-d2)*(d1-d2).
end loop
 
*print dist1.
loop ii = 1 to nr.
do if (in1(ii,2) > 0 or in1(ii,3) >0).
compute dist1(ii,2) = 0.
end if.
end loop.
*print dist1.
 
*print in1.
 
loop iii = 1 to nr.
 
 
compute sas1 = t(j1)*dist1.
compute ind12 = t(j1)*in1.

* Note.

do if ind12(1,2) <nr.
* print sas1.
* print in1.
* print sas1(1,2).
do if (sas1(1,2) > 0).
 
compute max1 = cmax(dist1).
 
* print max1.
	loop iv1 = 1 to nr.
		do if (abs(max1(1,2)-dist1(iv1,2) )< .000001).
		compute in1(iv1,2) = 1.
compute dist1(iv1,2) = 0.
		end if.	
end loop.
 
 
 
	compute max1 = cmax(dist1).
	loop iv2 = 1 to nr.
		do if (abs(max1(1,2)-dist1(iv2,2) )< .000001).
		compute in1(iv2,3) = 2.
compute dist1(iv2,2) = 0.
		end if.	
	end loop.
 
 
	
end if.
end if.
 
end loop.
*print dist1.
*print in1.
save in1/outfile = 'c:\in1.sav'.
 
end matrix.
 
 
GET
 FILE='c:\temp1.sav'.
MATCH FILES /FILE=*
 /FILE='C:\in1.sav'.
EXECUTE.

compute filter_$ = col2.
execute.
 
 SAVE OUTFILE='C:\good.sav'
  /COMPRESSED.

FILTER OFF.
USE ALL.
SELECT IF(filter_$ =0).
EXECUTE .
SAVE OUTFILE='C:\good1.sav' /drop col1 to filter_$  /COMPRESSED.
GET
  FILE='C:\good.sav'.
FILTER OFF.
USE ALL.
SELECT IF(filter_$=1).
EXECUTE .
SAVE OUTFILE='C:\good2.sav'
  /COMPRESSED.
GET
  FILE='C:\good.sav'.