Cluster arrays according to similarity of key values

The below script will compare a set of arrays according to similarities between their key's values. For example, if the first 4 keys values of an array are equal to another array's first 4 keys values, they are equal and consists a cluster. Here is my code:

<?php
$arrays = [ array('a'=>1, 'b'=>2, 'c'=>3, 'd'=>4), array('a'=>1, 'b'=>2, 'c'=>3, 'd'=>4), array('a'=>1, 'b'=>2, 'c'=>3, 'd'=>4), array('a'=>1, 'b'=>2, 'c'=>4, 'd'=>3), ];$result = [];

//get the keys of a sub-array that is inside $arrays, to be used later$keys = array_keys($arrays); for($i=0; $i < sizeof($arrays); $i++){$sa = array(); // to store similar arrays indexes

for($k=$i+1; $k < sizeof($arrays); $k++){$similar = false;

//compare the values of keys in the two arrays. Just compare the first 4 keys (as the user's desire)
for($j=0;$j < 4; $j++){ //check if the values are similar, if they are, assign$similar to true, and assign $j=3 to end the loop, (a bit of laziness here) ($similar = $arrays[$i][$keys[$j]] == $arrays[$k][$keys[$j]] ? true : false) ? null : ($j=3); } // check if the key (which represents an index in$arrays) is in $sa or not, if not, push it.$similar ? (in_array($i,$sa) ? null : array_push($sa,$i) && in_array($k,$sa) ? null : array_push($sa,$k)) : null;
//if $similar is true, make$i jumps to the $k index (saving time)$similar ? $i=$k : null;
}

//if $sa not empty, push it to$result
empty($sa) ? null : ($result[] = $sa); } /* // at this stage,$result includes all the similar arrays
// so we need another loop to push the unique arrays to $result // just check if an index of$arrays is in an sub-array of $result, if not, push it as an array of one record */ for($j=0; $j < sizeof($arrays); $j++){$f = false;
for($i=0;$i < sizeof($result);$i++){
in_array($j,$result[$i]) ?$f = true : null;
}
if(!$f){$sa = array();
array_push($sa,$j);
array_push($result,$sa);
}
}

If the result was as follows:

array(2) {
=> array(3) {
=> int(0)
=> int(1)
=> int(2)
},
=> array(1) {
=> int(3)
}
}

this means that $arrays has two clusters of sub-arrays, where$arrays, $arrays, and$arrays are similar (cluster 1), then \$arrays is unique (cluster 2).

Does this code have vulnerabilities? Could it be optimized?

• Are a number of existing array walking and sort routines that may help, such as usort – kph0x1 Jan 6 '16 at 14:20