No need to stop using employee names.

use strict;
use warnings;

use Data::Dumper qw( Dumper );

my %employees = (
   '01' => 'Joe',
   '02' => 'Ed',
   '03' => 'Sam',
   '04' => 'Wendy',
);

my @transactions = (
   'yadda 01 abc',
   'yadda 03 sss',
   'yadda 02 asdsdbc',
   'yadda 02 adfgbc',
   'yadda 01 abgxvc',
   'yadda 02 azdfbc',
   'yadda 04 azdfbc',
   'yadda 04 abdc',
);

my %transactions_by_employee;
for (@transactions) {
   my ($employee_number) = /\s(\d+)\s/;
   my $employee_name = $employees{$employee_number} // 'unregistered';
   push @{ $transactions_by_employee{$employee_name} }, $_;
}

print(Dumper(\%transactions_by_employee));
[download]

That said, it might be advantageous to only resolve the employee numbers during output. This would allow one to differentiate the different unregistered employees, for example.