view release on metacpan or  search on metacpan

lib/Math/LOESS/Prediction.pm  view on Meta::CPAN

    $alpha //= 0.01;

    unless ($alpha > 0 and $alpha < 1) {
        die "The alpha value should be between 0 and 1";
    }
    unless ($self->_obj->{se}) {
        die "Cannot compute confidence intervals without standard errors";
    }

    my $ci = Math::LOESS::_swig::confidence_intervals->new;
    Math::LOESS::_swig::pointwise($self->_obj, 1 - $alpha, $ci);   
    my $m = $self->_obj->{m};
    my $rslt =
      { map { $_ => Math::LOESS::_swig::darray_to_pdl( $ci->{$_}, $m ) }
          qw(fit upper lower) };
    Math::LOESS::_swig::pw_free_mem($ci);
    return $rslt;
}

1;

__END__

=pod

=encoding UTF-8

=head1 NAME

Math::LOESS::Prediction - Math::LOESS prediction and confidence intervals

=head1 VERSION

version 0.001000

=head1 DESCRIPTION

You normally don't need to construct object of this class yourself.
Instead you get the object from an L<Math::LOESS> object after its C<fit()>
method is called. 

=head1 ATTRIBUTES

=head2 values

loess values evaluated at newdata.

=head2 stderr

Estimates of the standard error on the estimated values.

=head2 residual_scale

Estimate of the scale of the residuals.

=head2 df

Degrees of freedom of the loess fit.

It is used with the t-distribution to compute pointwise confidence
intervals for the evaluated surface. It is obtained using the formula
C<(one_delta ** 2) / two_delta>

=head1 METHODS

=head2 confidence

    confidence($alpha=0.01)

Returns the confidence intervals for predicted values, as a hashref of
C<{ fit =E<gt> $fit, upper =E<gt> $upper, lower =E<gt> $lower }>,
where C<$fit>, C<$upper>, C<$lower> are piddles.

=head1 SEE ALSO

L<Math::LOESS>

=head1 AUTHOR

Stephan Loyd <sloyd@cpan.org>

=head1 COPYRIGHT AND LICENSE

This software is copyright (c) 2019-2023 by Stephan Loyd.

This is free software; you can redistribute it and/or modify it under
the same terms as the Perl 5 programming language system itself.

=cut