Bug#352454: libmath-numbercruncher-perl: Feeding the results of 'BestFit' into 'Predict' results in a wrong answer, in my case 'NAN'.

[Thomas Walter]

Package: libmath-numbercruncher-perl
Version: 5.00-1
Severity: important

Hello,
while using libmath-numbercruncher-perl I found the following issue.
Feeding the results of 'BestFit' into 'Predict' results in a wrong answer, in my case 'NAN'.

Example code snippet:

use v5.6.1;
use warnings;
use strict;

use Math::NumberCruncher;

my @array1 = (0, 1, 2, 3);
my @array2 = (5, 10, 12, 10);

my ($slope, $y_intercept) = Math::NumberCruncher::BestFit(\@array1, \@array2);

print "\$slope       = ", $slope, "\n";
print "\$y_intercept = ", $y_intercept, "\n";

my $proposed_x = 4;
my $predicted_value = Math::NumberCruncher::Predict( $slope, $y_intercept, 4);
#my $predicted_value = Math::NumberCruncher::Predict( 0.1, 3.1, 4); # runs fine
#my $predicted_value = Math::NumberCruncher::Predict( int($slope *1e8) / 1e8, int($y_intercept * 1e8)/1e8, $proposed_x); # OK too
print "\$proposed_x      = ", $proposed_x, "\n";
print "\$predicted_value = ", $predicted_value, "\n";

end of code snippet

As there is no wrong answer when using rounded values and when looking at the variables using 'perl -d' I assume the root of the issue is related to BigFloat numbers returned by 'BestFit'.  When feeding "normal" perl numbers into 'Predict' then everything is fine.

Kind Regards,
Thomas Walter
t.walter at nefkom.net


-- System Information
Debian Release: 3.0
Architecture: i386
Kernel: Linux seneca.org.eu 2.4.27 #2 Sun Apr 10 14:22:04 CEST 2005 i686
Locale: LANG=de, LC_CTYPE=C

Versions of packages libmath-numbercruncher-perl depends on:
ii  libmath-bigint-perl           1.56-1     Perl5 module for math. operation o
ii  perl                          5.6.1-8.9  Larry Wall's Practical Extraction