| line | stmt | bran | cond | sub | pod | time | code | 
| 1 |  |  |  |  |  |  | package Statistics::Hartigan; | 
| 2 |  |  |  |  |  |  |  | 
| 3 | 1 |  |  | 1 |  | 32580 | use 5.008005; | 
|  | 1 |  |  |  |  | 4 |  | 
|  | 1 |  |  |  |  | 50 |  | 
| 4 | 1 |  |  | 1 |  | 7 | use strict; | 
|  | 1 |  |  |  |  | 2 |  | 
|  | 1 |  |  |  |  | 37 |  | 
| 5 | 1 |  |  | 1 |  | 5 | use warnings; | 
|  | 1 |  |  |  |  | 8 |  | 
|  | 1 |  |  |  |  | 61 |  | 
| 6 |  |  |  |  |  |  |  | 
| 7 |  |  |  |  |  |  | require Exporter; | 
| 8 | 1 |  |  | 1 |  | 1155 | use AutoLoader qw(AUTOLOAD); | 
|  | 1 |  |  |  |  | 2234 |  | 
|  | 1 |  |  |  |  | 5 |  | 
| 9 |  |  |  |  |  |  |  | 
| 10 |  |  |  |  |  |  | our @ISA = qw(Exporter); | 
| 11 |  |  |  |  |  |  |  | 
| 12 |  |  |  |  |  |  | our @EXPORT = qw( hartigan ); | 
| 13 |  |  |  |  |  |  |  | 
| 14 |  |  |  |  |  |  | our $VERSION = '0.01'; | 
| 15 |  |  |  |  |  |  |  | 
| 16 |  |  |  |  |  |  | # global variable | 
| 17 |  |  |  |  |  |  | my @H = (); | 
| 18 |  |  |  |  |  |  | my @d = (); | 
| 19 |  |  |  |  |  |  | my @W = (); | 
| 20 |  |  |  |  |  |  | my $rcnt = 0; | 
| 21 |  |  |  |  |  |  |  | 
| 22 |  |  |  |  |  |  | sub hartigan | 
| 23 |  |  |  |  |  |  | { | 
| 24 |  |  |  |  |  |  | # Input params | 
| 25 | 0 |  |  | 0 | 0 |  | my $matrixfile = shift; | 
| 26 | 0 |  |  |  |  |  | my $clustmtd = shift; | 
| 27 | 0 |  |  |  |  |  | my $K = shift; | 
| 28 | 0 |  |  |  |  |  | my $threshold = shift; | 
| 29 |  |  |  |  |  |  |  | 
| 30 | 0 |  |  |  |  |  | my $i = 0; | 
| 31 | 0 |  |  |  |  |  | my $j = 0; | 
| 32 |  |  |  |  |  |  |  | 
| 33 |  |  |  |  |  |  | # Read the matrix file into a 2 dimensional array. | 
| 34 | 0 |  |  |  |  |  | my @inpmat = (); | 
| 35 | 0 | 0 |  |  |  |  | open(INP,"<$matrixfile") || die "Error opening input matrix file!"; | 
| 36 |  |  |  |  |  |  |  | 
| 37 |  |  |  |  |  |  | # Extract the number of rows from the first line in the file. | 
| 38 | 0 |  |  |  |  |  | my $ccnt = 0; | 
| 39 | 0 |  |  |  |  |  | my $line; | 
| 40 |  |  |  |  |  |  |  | 
| 41 | 0 |  |  |  |  |  | $line = ; | 
| 42 | 0 |  |  |  |  |  | chomp($line); | 
| 43 | 0 |  |  |  |  |  | $line=~s/\s+/ /; | 
| 44 |  |  |  |  |  |  |  | 
| 45 | 0 |  |  |  |  |  | ($rcnt,$ccnt) = split(/\s+/,$line); | 
| 46 |  |  |  |  |  |  |  | 
| 47 |  |  |  |  |  |  | # Not a valid condition: | 
| 48 |  |  |  |  |  |  | # If maximum number of clusters requested (k) is greater than the | 
| 49 |  |  |  |  |  |  | # number of observations. | 
| 50 | 0 | 0 |  |  |  |  | if($K > $rcnt) | 
| 51 |  |  |  |  |  |  | { | 
| 52 | 0 |  |  |  |  |  | print STDERR "The K value ($K) cannot be greater than the number of observations present in the input data ($rcnt). \n"; | 
| 53 | 0 |  |  |  |  |  | exit 1; | 
| 54 |  |  |  |  |  |  | } | 
| 55 |  |  |  |  |  |  |  | 
| 56 |  |  |  |  |  |  | # Copy the complete matrix to a 2D array | 
| 57 | 0 |  |  |  |  |  | while() | 
| 58 |  |  |  |  |  |  | { | 
| 59 |  |  |  |  |  |  | # remove the newline at the end of the input line | 
| 60 | 0 |  |  |  |  |  | chomp; | 
| 61 |  |  |  |  |  |  |  | 
| 62 |  |  |  |  |  |  | # skip empty lines | 
| 63 | 0 | 0 |  |  |  |  | if(m/^\s*\s*\s*$/) | 
| 64 |  |  |  |  |  |  | { | 
| 65 | 0 |  |  |  |  |  | next; | 
| 66 |  |  |  |  |  |  | } | 
| 67 |  |  |  |  |  |  |  | 
| 68 |  |  |  |  |  |  | # remove leading white spaces | 
| 69 | 0 |  |  |  |  |  | s/^\s+//; | 
| 70 |  |  |  |  |  |  |  | 
| 71 |  |  |  |  |  |  | # seperate individual values in a line | 
| 72 | 0 |  |  |  |  |  | my @tmp = (); | 
| 73 | 0 |  |  |  |  |  | @tmp = split(/\s+/); | 
| 74 |  |  |  |  |  |  |  | 
| 75 |  |  |  |  |  |  | # populate them into the 2D matrix | 
| 76 | 0 |  |  |  |  |  | push @inpmat, [ @tmp ]; | 
| 77 |  |  |  |  |  |  | } | 
| 78 |  |  |  |  |  |  |  | 
| 79 | 0 |  |  |  |  |  | close INP; | 
| 80 |  |  |  |  |  |  |  | 
| 81 | 0 |  |  |  |  |  | my @row1 = (); | 
| 82 | 0 |  |  |  |  |  | my @row2 = (); | 
| 83 |  |  |  |  |  |  |  | 
| 84 | 0 |  |  |  |  |  | my %hash = (); | 
| 85 |  |  |  |  |  |  |  | 
| 86 |  |  |  |  |  |  | # Calculate all possible unique pairwise distances between the vectors | 
| 87 | 0 |  |  |  |  |  | for($i = 0; $i < $rcnt; $i++) | 
| 88 |  |  |  |  |  |  | { | 
| 89 |  |  |  |  |  |  | # for all the rows in the cluster | 
| 90 | 0 |  |  |  |  |  | for($j = $i+1; $j < $rcnt; $j++) | 
| 91 |  |  |  |  |  |  | { | 
| 92 | 0 |  |  |  |  |  | @row1 = @{$inpmat[$i]}; | 
|  | 0 |  |  |  |  |  |  | 
| 93 | 0 |  |  |  |  |  | @row2 = @{$inpmat[$j]}; | 
|  | 0 |  |  |  |  |  |  | 
| 94 | 0 |  |  |  |  |  | $d[$i][$j] = &dist_euclidean_sqr(\@row1, \@row2); | 
| 95 |  |  |  |  |  |  | } | 
| 96 | 0 |  |  |  |  |  | $hash{0} .= "$i "; | 
| 97 |  |  |  |  |  |  | } | 
| 98 |  |  |  |  |  |  |  | 
| 99 | 0 |  |  |  |  |  | $hash{0} = substr($hash{0},0,length($hash{0})-1); | 
| 100 | 0 |  |  |  |  |  | $W[1] = &WGSS(\%hash); | 
| 101 |  |  |  |  |  |  |  | 
| 102 | 0 |  |  |  |  |  | my $k = 0; | 
| 103 | 0 |  |  |  |  |  | my $flag = 0; | 
| 104 |  |  |  |  |  |  |  | 
| 105 |  |  |  |  |  |  | # For each K | 
| 106 | 0 |  |  |  |  |  | for($k=1; $k<$K; $k++) | 
| 107 |  |  |  |  |  |  | { | 
| 108 | 0 |  |  |  |  |  | my $lineNo = 0; | 
| 109 | 0 |  |  |  |  |  | my %hash = (); | 
| 110 |  |  |  |  |  |  |  | 
| 111 |  |  |  |  |  |  | # Cluster the input dataset into k+1 clusters | 
| 112 | 0 |  |  |  |  |  | my $status = 0; | 
| 113 | 0 |  |  |  |  |  | my $next_k = $k + 1; | 
| 114 | 0 |  |  |  |  |  | $status = system("vcluster --clmethod $clustmtd $matrixfile $next_k >& tmpfile"); | 
| 115 | 0 | 0 |  |  |  |  | die "Error running vcluster \n" unless $status==0; | 
| 116 |  |  |  |  |  |  |  | 
| 117 |  |  |  |  |  |  | # read the clustering output file | 
| 118 | 0 | 0 |  |  |  |  | open(CO,"<$matrixfile.clustering.$next_k") || die "Error opening clustering output file."; | 
| 119 |  |  |  |  |  |  |  | 
| 120 | 0 |  |  |  |  |  | my $clust = 0; | 
| 121 | 0 |  |  |  |  |  | while($clust = ) | 
| 122 |  |  |  |  |  |  | { | 
| 123 |  |  |  |  |  |  | # hash on the cluster# and append the observation# | 
| 124 | 0 |  |  |  |  |  | chomp($clust); | 
| 125 | 0 | 0 |  |  |  |  | if(exists $hash{$clust}) | 
| 126 |  |  |  |  |  |  | { | 
| 127 | 0 |  |  |  |  |  | $hash{$clust} .= " $lineNo"; | 
| 128 |  |  |  |  |  |  | } | 
| 129 |  |  |  |  |  |  | else | 
| 130 |  |  |  |  |  |  | { | 
| 131 | 0 |  |  |  |  |  | $hash{$clust} = $lineNo; | 
| 132 |  |  |  |  |  |  | } | 
| 133 |  |  |  |  |  |  |  | 
| 134 |  |  |  |  |  |  | # increment the line number | 
| 135 | 0 |  |  |  |  |  | $lineNo++; | 
| 136 |  |  |  |  |  |  | } | 
| 137 |  |  |  |  |  |  |  | 
| 138 | 0 |  |  |  |  |  | close CO; | 
| 139 |  |  |  |  |  |  |  | 
| 140 |  |  |  |  |  |  | # Calculate the "Within Cluster Sum of Squared W(k+1) | 
| 141 |  |  |  |  |  |  | # for given matrix and k+1 value. | 
| 142 | 0 |  |  |  |  |  | $W[$next_k] = &WGSS(\%hash); | 
| 143 |  |  |  |  |  |  |  | 
| 144 | 0 |  |  |  |  |  | unlink "$matrixfile.clustering.$next_k", "tmpfile","$matrixfile.tree"; | 
| 145 |  |  |  |  |  |  |  | 
| 146 | 0 | 0 |  |  |  |  | if($W[$next_k] == 0) | 
| 147 |  |  |  |  |  |  | { | 
| 148 | 0 |  |  |  |  |  | print $next_k . "\n"; | 
| 149 | 0 |  |  |  |  |  | $flag = 1; | 
| 150 | 0 |  |  |  |  |  | last; | 
| 151 |  |  |  |  |  |  | } | 
| 152 |  |  |  |  |  |  | else | 
| 153 |  |  |  |  |  |  | { | 
| 154 | 0 |  |  |  |  |  | $H[$k] = ( $W[$k]/$W[$next_k] - 1 ) * ( $rcnt - $k - 1 ); | 
| 155 | 0 |  |  |  |  |  | $H[$k] = sprintf("%.4f",$H[$k]); | 
| 156 |  |  |  |  |  |  | } | 
| 157 |  |  |  |  |  |  |  | 
| 158 | 0 | 0 |  |  |  |  | if($H[$k] <= $threshold) | 
| 159 |  |  |  |  |  |  | { | 
| 160 | 0 |  |  |  |  |  | print $k . "\n"; | 
| 161 | 0 |  |  |  |  |  | $flag = 1; | 
| 162 | 0 |  |  |  |  |  | last; | 
| 163 |  |  |  |  |  |  | } | 
| 164 |  |  |  |  |  |  | } | 
| 165 |  |  |  |  |  |  |  | 
| 166 | 0 | 0 |  |  |  |  | if(!$flag) | 
| 167 |  |  |  |  |  |  | { | 
| 168 | 0 |  |  |  |  |  | print "NAN\n"; | 
| 169 |  |  |  |  |  |  | } | 
| 170 |  |  |  |  |  |  | } | 
| 171 |  |  |  |  |  |  |  | 
| 172 |  |  |  |  |  |  | sub WGSS | 
| 173 |  |  |  |  |  |  | { | 
| 174 |  |  |  |  |  |  | # Input arguments | 
| 175 | 0 |  |  | 0 | 0 |  | my %clustout = %{(shift)}; | 
|  | 0 |  |  |  |  |  |  | 
| 176 |  |  |  |  |  |  |  | 
| 177 |  |  |  |  |  |  | # Local variables | 
| 178 | 0 |  |  |  |  |  | my $i; | 
| 179 |  |  |  |  |  |  | my $j; | 
| 180 | 0 |  |  |  |  |  | my @rownum; | 
| 181 | 0 |  |  |  |  |  | my $key; | 
| 182 | 0 |  |  |  |  |  | my $row1; | 
| 183 | 0 |  |  |  |  |  | my $row2; | 
| 184 | 0 |  |  |  |  |  | my @D = (); | 
| 185 | 0 |  |  |  |  |  | my $W = 0; | 
| 186 |  |  |  |  |  |  |  | 
| 187 |  |  |  |  |  |  | # For each cluster | 
| 188 | 0 |  |  |  |  |  | foreach $key (sort keys %clustout) | 
| 189 |  |  |  |  |  |  | { | 
| 190 | 0 |  |  |  |  |  | $D[$key] = 0; | 
| 191 |  |  |  |  |  |  |  | 
| 192 | 0 |  |  |  |  |  | @rownum = split(/\s+/,$clustout{$key}); | 
| 193 |  |  |  |  |  |  |  | 
| 194 |  |  |  |  |  |  | # for each instance in the cluster | 
| 195 | 0 |  |  |  |  |  | for($i = 0; $i < $#rownum; $i++) | 
| 196 |  |  |  |  |  |  | { | 
| 197 |  |  |  |  |  |  | # for all the rows in the cluster | 
| 198 | 0 |  |  |  |  |  | for($j = $i+1; $j <= $#rownum; $j++) | 
| 199 |  |  |  |  |  |  | { | 
| 200 |  |  |  |  |  |  | # find the distance between the 2 rows of the matrix. | 
| 201 | 0 |  |  |  |  |  | $row1 = $rownum[$i]; | 
| 202 | 0 |  |  |  |  |  | $row2 = $rownum[$j]; | 
| 203 |  |  |  |  |  |  |  | 
| 204 |  |  |  |  |  |  | # store the Dr value | 
| 205 | 0 | 0 |  |  |  |  | if(exists $d[$row1][$row2]) | 
| 206 |  |  |  |  |  |  | { | 
| 207 | 0 |  |  |  |  |  | $D[$key] += $d[$row1][$row2]; | 
| 208 |  |  |  |  |  |  | } | 
| 209 |  |  |  |  |  |  | else | 
| 210 |  |  |  |  |  |  | { | 
| 211 | 0 |  |  |  |  |  | $D[$key] += $d[$row2][$row1]; | 
| 212 |  |  |  |  |  |  | } | 
| 213 |  |  |  |  |  |  | } | 
| 214 |  |  |  |  |  |  | } | 
| 215 |  |  |  |  |  |  |  | 
| 216 | 0 |  |  |  |  |  | $W += ( $#rownum - 1 ) * $D[$key]; | 
| 217 |  |  |  |  |  |  | } | 
| 218 |  |  |  |  |  |  |  | 
| 219 | 0 |  |  |  |  |  | $W = $W/2; | 
| 220 |  |  |  |  |  |  |  | 
| 221 | 0 |  |  |  |  |  | return $W; | 
| 222 |  |  |  |  |  |  | } | 
| 223 |  |  |  |  |  |  |  | 
| 224 |  |  |  |  |  |  | sub dist_euclidean_sqr | 
| 225 |  |  |  |  |  |  | { | 
| 226 |  |  |  |  |  |  | # arguments | 
| 227 | 0 |  |  | 0 | 0 |  | my @i = @{(shift)}; | 
|  | 0 |  |  |  |  |  |  | 
| 228 | 0 |  |  |  |  |  | my @j = @{(shift)}; | 
|  | 0 |  |  |  |  |  |  | 
| 229 |  |  |  |  |  |  |  | 
| 230 |  |  |  |  |  |  | # local variables | 
| 231 | 0 |  |  |  |  |  | my $a; | 
| 232 | 0 |  |  |  |  |  | my $dist = 0; | 
| 233 | 0 |  |  |  |  |  | my $retvalue = 0; | 
| 234 |  |  |  |  |  |  |  | 
| 235 |  |  |  |  |  |  | # Squared Euclidean measure | 
| 236 |  |  |  |  |  |  | # summation on all j (xij - xi'j)^2 where i, i' are the rows indicies. | 
| 237 | 0 |  |  |  |  |  | for $a (0 .. $#i) | 
| 238 |  |  |  |  |  |  | { | 
| 239 | 0 |  |  |  |  |  | $dist += (($i[$a] - $j[$a])**2); | 
| 240 |  |  |  |  |  |  | } | 
| 241 |  |  |  |  |  |  |  | 
| 242 | 0 |  |  |  |  |  | $retvalue = sprintf("%.4f",$dist); | 
| 243 | 0 |  |  |  |  |  | return $retvalue; | 
| 244 |  |  |  |  |  |  | } | 
| 245 |  |  |  |  |  |  |  | 
| 246 |  |  |  |  |  |  |  | 
| 247 |  |  |  |  |  |  | 1; | 
| 248 |  |  |  |  |  |  | __END__ |