«

For a market basket analysis, I'd like to traverse a hash structure
containing entries like:

my %list;
$list{"c1"} = "milk%cheese%coca cola";
$list{"c2"} = "coca cola";
$list{"c3"} = "beef";
....
$list{"cn-1"} = "apple%cheese";
$list{"cn"} = "beef%milk";

so the knowledge derived may look like:
c1 is associated with c2, .., cn-1, cn
c3 is associated with cn
....

i.e. the total number of checking is "n + n-1 + ... + 2 + 1" .
While this can be easily achieved if the structure is array, the following
looks clumsy... any better implementation?

while (%list) {
$assign = 1;
foreach $key (keys %list) {
if ($assign == 1) {
$head = $key;
$assign = 0;
next;
}
if ($list{$head} =~ /$list{$key}/ or $list{$key} =~
/$list{$head}/ ) {
print "$list{$head}\t$list{$key}\t1\n";
}
}
undef $list{$head};
}

>>>>> "HL" == Henry Law <> writes:

HL> On 29/08/11 05:01, ela wrote:
>> For a market basket analysis, I'd like to traverse a hash structure
>> containing entries like:
>>
>> my %list;
>> $list{"c1"} = "milk%cheese%coca cola";

HL> Just by the way, is this real code? Surely Perl would have tried to
HL> parse %cheese etc as the names of hash variables? Confusing at best.

confusing maybe but hashes don't interpolate in strings.

uri

--
Uri Guttman -- uri AT perlhunter DOT com --- http://www.perlhunter.com --
------------ Perl Developer Recruiting and Placement Services -------------
----- Perl Code Review, Architecture, Development, Training, Support -------

On Mon, 29 Aug 2011 12:01:40 +0800, "ela" <> wrote:

>For a market basket analysis, I'd like to traverse a hash structure
>containing entries like:
>
>my %list;
>$list{"c1"} = "milk%cheese%coca cola";
>$list{"c2"} = "coca cola";
>$list{"c3"} = "beef";
>...
>$list{"cn-1"} = "apple%cheese";
>$list{"cn"} = "beef%milk";
>
>so the knowledge derived may look like:
>c1 is associated with c2, .., cn-1, cn
>c3 is associated with cn
>...
>
>i.e. the total number of checking is "n + n-1 + ... + 2 + 1" .
>While this can be easily achieved if the structure is array, the following
>looks clumsy... any better implementation?
>
>while (%list) {
> $assign = 1;
> foreach $key (keys %list) {
> if ($assign == 1) {
> $head = $key;
> $assign = 0;
> next;
> }
> if ($list{$head} =~ /$list{$key}/ or $list{$key} =~
>/$list{$head}/ ) {
> print "$list{$head}\t$list{$key}\t1\n";
> }
> }
> undef $list{$head};
>}
>

Its still being treated as a list in the foreach loop.
In your current implementation, the %list keys never
change so you sit in the while() loop forever, and
$head is assigned the same $key every time, doing the
same operation indefinitely and generating uninitialized
value warnings in the regex.

change this:
> undef $list{$head};

to this:
delete $list{$head};

As a final note, the target of a regular expression
doesen't transparently translate into a regular expression itself.

-sln

<> wrote in message
news...
> Its still being treated as a list in the foreach loop.
> In your current implementation, the %list keys never
> change so you sit in the while() loop forever, and
> $head is assigned the same $key every time, doing the
> same operation indefinitely and generating uninitialized
> value warnings in the regex.
>
> change this:
>> undef $list{$head};
>
> to this:
> delete $list{$head};
>
> As a final note, the target of a regular expression
> doesen't transparently translate into a regular expression itself.
>
> -sln

Thanks sln. I hope you have already got your favorite job. Thanks again