Hashes are (by definition) unordered so when you call values you're getting the values in a (pseudo)random order. If you want them back in the order of the keys you either need to store the keys off into an array as they come in, or use keys on your hash (which will return items in the same order as the corresponding values; alternately sort the keys and iterate over that to pull out and print the corresponding value).

Update: Just to expand merging the suggestion above roughly something like this.

my %hash;
my @key_order;
while( defined( my $line = <> ) ) {
  my ($first, $third) = (split ' ', $line)[0, 2];
  $hash{ $first } = $third;
  push @key_order, $first;
}

## In the order they appeared . . .
for my $key (@key_order) {
  say $hash{ $key };
}

## Or in their random ordering but with the match . . .
for my $key ( keys %hash ) {
  say qq{$key => $hash{ $key }};
}
[download]

Now i am counting the values in the third column i.e how many up words how many down words and how many NA. For that i wrote the following program:

@values = values(%hash);

for $element (@values){
 if ($element =~ /up/){
    $sumA++; 
}if(($element =~ /down/)){
    $sumB++
}if(($element =~ /NA/)){
    $sumC++
}
}
print "Number of up is: $sumA\n";
print "Number of down is: $sumB\n";
print "Number of NA is: $sumC\n";
[download]

it is doing a great job but when i add a same row in the last of the file it doesn't count its up, down or NA word. why is that?

Presuming you mean that you're adding a row which duplicates the key and value from another row you're not going to change your hash. If you've seen a key "foo" with a value "up", setting $hash{ "foo" } = "up" again doesn't really change anything in %hash. Hashes are unordered and maintain a single value for each key slot. If you need the count of rows with a particular value then count them as you read things.

my %counts_per_value;

  [... inside your while loop ...]
  $hash{ $first } = $third;
  $counts_per_value{ $third }++;
  [...]

for my $value ( sort keys %counts_per_value ) {
  say qq{$value => $counts_per_value{ $value };
}
[download]

The cake is a lie.
The cake is a lie.
The cake is a lie.

Because the hash contains each key just once.
map{substr$_->[0],$_->[1]||0,1}[\*||{},3],[[]],[ref qr-1,-,-1],[{}],[sub{}^*ARGV,3]

You seem to misunderstand or lack understanding of the behavior of hashes, or associative arrays to use their formal name. Please see the discussions of keys, values and each in the Perl documentation.

---

Give a man a fish:  <%-{-{-{-<