Tim Bunce wrote at Thu, 11 Jul 2002 17:56:35 +0200:
> Why 'Cross'? Makes me think of 'pivot' cross-tabulations, which are a different
>thing.
I hadn't the association to pivot cross-tabulation.
I simply translated the german word "Kreuztabelle" directly into english.
I think the name comes as every data cell influence
its row sum, its col sum and finally the total table sum,
what gives a picture like that (where the XXX are modified)
Col1 Col2 Col3 Sum
Row1 ??? XXX ??? XXX
Row2 ??? ??? ??? ???
Row3 ??? ??? ??? ???
Sum ??? XXX ??? XXX
Perhaps it doesn't like very much like a cross,
but our calculation seems to go along "crossings".
However,
if the name isn't intuitive enough,
we really should find a better one.
>
> See http://search.cpan.org/search?mode=doc&query=pivot
> There's already Data::Pivot, Data::Pivoter,
> and Data::Xtab (which sounds close to yours even if it's spelt differently :)
I didn't found them on my first review,
as I hadn't the association to pivoting
(as I don't rotate or loop really something :-)),
and read Xtab as eXtending tabbing.
(By a table I would have expected an uppercase T)
The both Pivot modules are mainly for creating some sensful
output of tables,
but I want to calculate with the tables.
It's similar to Data::Xtab.
What I primarly want is some statistical calculations.
Let's think e.g. to a population count.
Let's say we want to create a table showing the population
in the U.S.A. regarding to the towns
'New York', 'Chicago', 'L.A.', 'San Francisco', 'Houston'
and the sex
'female', 'male'.
Let's think I've got a big file with information of the population,
e.g. the phone books.
Then I'd like to write a pseudocode like
my @towns = ('New York', 'Chicago', 'L.A.', 'San Francisco', 'Houston');
my @sex = qw/female male/;
my $population = Data::CrossTable->new(rows => \@sex, cols => \@towns);
while (<FILE>) {
my ($town, $sex) = /...../;
$population->data($sex,$town)++;
}
where the populations of each town, each sex and the total population
are updated in the background.
Finally we could say,
e.g. there are $population->rowresult('female') women in the big cities.
Furthermore I'd like to calculate with the CrossTables.
E.g. the percentuage of inhabitants could be calculated with:
my $perc_population = $population / $population->totalresult();
or if we would like to have the more readable 12% reading than the 0.12,
we could write something like
my $perc100_population = $perc_population->change( sub { $_ *= 100 } );
Another way is that we have built the tax payings for the big cities.
Then we could write
my $tax_per_inhabitent = $tax_per_city_and_sex / $population;
Of course,
a nice output is still important,
but I really want to avoid me to always write
foreach (...) { $_ *= 100 for ( ... ) }
lines.
>
> How about Data::TableAutoSum?
I don't know.
I'd prefer to work also with other associative and distribute operations.
E.g., we could find the maximum living people in a town by:
$population->op(sub {max @_});
And now $population->totalresult() contains the maximum living people.
More sensful could be the multiplication when working with probabilities or
something like an average (allthough it isn't distributive).
So I don't really like the association to Sum as Addition.
Summarize could be O.K.,
but Data::TableAutoSummarize would sound like I'll make a summerizing of every table.
Greetings,
Janek
