That was actually "correct", 12/5=2.4, and 22/5 = 4.4.

Only digits before commas are converted.

use strict;
use warnings;

while (my $line = <DATA>) {
    $line =~ s/,//g;
    my @n = split(/\s+/, $line);
    my $sum = 0;
    for (@n) {$sum += $_};
    print $sum / ($#n + 1), "\n";
}

__DATA__
2,800  2,900 3,000  3,000  2,988 
5,000  5,495 4,999  4,995  4,900
[download]

This gives:

2937.6
5077.8
[download]

Instead of $line =~ s/,//g; to remove commas, I would use a more generalized regex (but on each element, not on each line):

$elem =~ s/[^\d+-e.]//g;
[download]

This will remove any char that is not a digit, a +, a -, an 'e' (for numbers like 2e+31), or a point (.).

Or, even better, use Number::Format:

use Number::Format;
$elem = unformat_number($elem);
[download]

This is especially useful if an array is used instead of the $se# notation. You'd access each one through $se[#] instead, and you could do this:

for (@se) { $_ = $unformat_number($_) }
[download]

<-radiant.matrix->
A collection of thoughts and links from the minds of geeks
The Code that can be seen is not the true Code
"In any sufficiently large group of people, most are idiots" - Kaa's Law

perl -le 'print "2,300" + "3,500"'

Output:

5

It doesn't really just make a guess. It grabs all the numeric value it can find from the front of the string, and truncates anything past that. You can see the effect in action:

my @strings = qw(
    2,300
    2.54xyz123
    2e32abc3
    2.34,23
);

for (@strings) {
    print "$_ => ", 0+$_, "\n";
}
[download]

This prints:

2,300 => 2
2.54xyz123 => 2.54
2e32abc3 => 2e+32
2.34,23 => 2.34
[download]

You have shown a tiny part of your code but not, I suspect, the part likely to contain the error. How about posting a chunk of code that can be run and that shows the problem. For strong preference use <DATA> to read some sample data from a __DATA__ section. It would also help a lot if your output columns lined up with the headings!

It is highly likely that your problem is in code that splits or otherwise extracts the data from your input stream.

There does seem to be a correlation between input containing commas and bogus averages. Is that a hint?

1 5,000 5,495 4,999 4,995 4,900 4,999 4.4 5,495

I won't repeat your problem with commas but, have you noticed that in this line, min is wrong?