Remove array positions that have the common name

With your current code, for the purpose stated, you can do this:

$arr = array(
  'id'              => '147',
  'nome'            =>'João',
  'email'           => '[email protected]',
  'data_nascimento' => '05/01/1987',
  'numero_1'        => '1',
  'rua_1'           => 'rua 1', 
  'numero_2'        => '2',
  'rua_2'           => 'rua 2',
  'submit'          => 'Salvar' 
);

$newArr = array();

foreach ($arr as $k => $v)                    // por cada entrada na matriz
    if (strpos($k, "numero_") === 0)          // se começar por "numero_"
        $newArr[$k] = $v;                     // envia para a nova matriz

The result will be:

array(2) {
  ["numero_1"]=>
  string(1) "1"
  ["numero_2"]=>
  string(1) "2"
}

See example on Ideone .

I thought the intent is to detect any duplicity in the name before the underscore ( underscore ) and not just the "number". And example does not match the description of the question.

So I came up with this code:

$arr = array(
  'id'              => '147',
  'nome'            =>'João',
  'email'           => '[email protected]',
  'data_nascimento' => '05/01/1987',
  'numero_1'        => '1',
  'rua_1'           => 'rua 1', 
  'numero_2'        => '2',
  'rua_2'           => 'rua 2',
  'submit'          => 'Salvar' 
);
$newArr = array();
foreach ($arr as $k => $v) {
    $partes = explode("_", $k);
    if (count($partes) > 1) {
        $repetido = 0;
        foreach ($arr as $k2 => $v2) {
            if (strpos($k2, $partes[0]."_") === 0) {
                $repetido++;
            }
        }
        if ($repetido > 1)
            $newArr[$k] = $v;
    }
}
var_dump($newArr);

See running on ideone . And no Coding Ground . Also I placed GitHub for future reference .

I think you can optimize something but the basis is this.

Result:

array(4) {
  ["numero_1"]=>
  string(1) "1"
  ["rua_1"]=>
  string(5) "rua 1"
  ["numero_2"]=>
  string(1) "2"
  ["rua_2"]=>
  string(5) "rua 2"
}

If I understand correctly the purpose is to transfer from the input array all elements that have the same prefix characterized by an expression followed by an underscore .

So, I propose the scheme below:

$filtered = array();

array_walk(

    $arr,

    function( $current, $offset ) use( $arr, &$filtered ) {

         preg_match( '/(\w+_\d+)/', $offset, $matches );

        if( count( $matches ) > 0 && array_key_exists( $matches[ 1 ], $arr ) ) {

            $filtered[ $matches[ 1 ] ] = $current;
        }
    }
);

Using the input array with the same name you have today ( $ arr ) the resulting array $ filtered will have the following values:

array(
  'numero_1' => '1',
  'rua_1' => 'rua 1',
  'numero_2' => '2',
  'rua_2' => 'rua 2'
)

Your second question can be answered simply by computing the difference between the two arrays, original and filtered:

$diff = array_diff_assoc( $arr, $filtered );

The $ diff array has the following values:

array(
  'id' => '147',
  'nome' => 'João',
  'email' => '[email protected]',
  'data_nascimento' => '05/01/1987',
  'submit' => 'Salvar'
)

This approach has the following advantages:

• Do not depend on a fixed expression, that is, as long as all similarities follow the same format prefix_number , it will work in the entire array.
• It is fast, because it dispenses calculations, complex conditions and nested loops. So even if large arrays are bad enough for your Application, this procedure is not going to be your bottleneck.

My contribution, mixing some of the answers already given:

$arr = array(
  'id'              => '147',
  'nome'            =>'João',
  'email'           => '[email protected]',
  'data_nascimento' => '05/01/1987',
  'numero_1'        => '1',
  'rua_1'           => 'rua 1', 
  'numero_2'        => '2',
  'rua_2'           => 'rua 2',
  'submit'          => 'Salvar' 
);

$padrao = "/[\w]+\_[\d]+/i";
$chaves = array_keys($arr);

foreach ($arr as $chaveAtual => $valor) {
    preg_match($padrao, $chaveAtual, $encontrados);
    $padraoEncontrado = count($encontrados);
    $padraoSeRepete = (int) preg_grep($padrao,$chaves);
    if ((!$padraoEncontrado) || (!$padraoSeRepete)) {
        unset($arr[$chaveAtual]);
    }
}

var_dump($arr);

/*
Resultado:
array(4) {  
  ["numero_1"]=>
  string(1) "1"
  ["rua_1"]=>
  string(5) "rua 1"
  ["numero_2"]=>
  string(1) "2"
  ["rua_2"]=>
  string(5) "rua 2"
}
*/