* %ARI - A SAS Macro written by von Borries (2008) PhD Thesis, Kansas State University to
* calculate the Adjusted Rand statistic (Hubert & Arabie, 1985);
* Coding was updated from Fisher and Hoffman (1988) to work in SAS Ver 9x;
* Additional modification by Bowley (2008) for direct listing of result;
* Two options for handling missing clusters:
* I) Each missing value becomes its own cluster. MISSING=SEPARATE ;
* 2) M*N table ---> (M+I)*(N+I). MISSING=COMBINED;

%macro tab(data=,missing=);
proc iml;
start a2;
do i=1 to nrow(z);
t[z[i,1],z[i,2]] = z[i,3];
end;
finish;
start separate;
do k = 1 to nrow(z);
if z0[k,1] = . then if z0[k,2] = . then do;
in = i:i+z0[k,3]-1;
jn = j:j+z0[k,3]-1;
z = z // (in || jn || j(z0[k,3],1,1));
i = i + z0[k,3];
j = j + z0[k,3];
nomiss = 0;
end;
else do;
in = i:i+z0[k,3]-1;
z = z // (in || j(z0[k,3],1,z0[k,2]) || j(z0[k,3],1,1));
i = i + z0[k,3];
nomiss = 0;
end;
else if z[k,2] = . then do;
jn = j:j+z0[k,3]-1;
z = z // (j(z0[k,3],1,z0[k,1]) || jn || j(z0[k,3],1,1));
j = j + z0[k,3];
nomiss = 0;
end;
else keep = keep || k;
end;
if nomiss = 0 then z = z[keep || (nrow(z0)+1:nrow(z)),];
z0 = z;
z[rank(z[,1]),] = z0;
finish;
start combined;
x0 = loc(z[,1] = .);
y0 = loc(z[,2] = .);
if ncol(x0) > 0 then z[x0,1] = i;
if ncol(y0) > 0 then z[y0,2] = j;
if ((ncol(x0) > 0) | (ncol(y0) > 0)) then nomiss = 0;
finish;
use tabari;
read all into z0;
i = max(z0[,1])+1;
j = max(z0[,2])+1;
z = z0;
nomiss = 1;
if ((&missing = 'separate') | (&missing = '0')) then call separate;
if ((&missing = 'combined') | (&missing = '1')) then call combined;
t = j(max(z[,1]),max(z[,2]),0);
call a2;
cm = t[+,];
sscm = ssq(cm);
rm = t[,+];
ssrm = ssq(rm);
tot = rm[+];
sst = ssq(t);
nn = tot * tot;
n2 = 0.5 * tot * (tot - 1);
n0 = -0.5 * (ssrm + sscm) + n2;
ns = sst + n0;
nc = ssrm * sscm / nn + n0 + (nn - ssrm) * (nn - sscm) / (nn * (tot - 1));
adjrand = (ns - nc) / (n2 - nc);
TotalNumber=tot;
print nomiss sscm ssrm tot sst;
print TotalNumber;
res = nomiss || sscm || ssrm || tot || sst || adrand;
print adjrand;
create adjrand var{adrand};
append from adjrand;
quit;
%mend tab;
%macro freq(data=,x=,y=) ;*/ store des='ARI - FREQ'; ;
proc freq data = &data;
tables &x * &y / out = tabari(rename=(&x = x &y = y) drop=percent) noprint;
run;
%mend freq;
%macro ari(data,x,y) ;*/ store des='ARI - MAIN';
%let missing='separate';
data &data;
set &data;
z = &x + 1;
k = &y + 1;
run;
%freq(data=&data,x=z,y=k);
%tab(data=&data,missing=&missing);
data &data(drop=z k);
set &data;
run;
%mend ari;