lifestyle-csv2vw

#!/usr/bin/perl -w
#
# Generate a VW training-set from our raw data CSV
#
use Scalar::Util qw(looks_like_number);

my $SepPat = qr{(?:\s*,\s*|\s+)};
my $Interactive = -t STDOUT;

# Default days-window
my $NDays = 1;

sub process_args() {
    for my $arg (@ARGV) {
        if (! -f $arg and $arg =~ /^\d$/) {
            $NDays = $arg;
        } else {
            push(@files, $arg);
        }
    }
    @ARGV = @files;
}

sub process_input() {
    my $prev_weight;
    my @daily = ();

    while (<>) {
        # Skip comments or header-line
        next if (/^[#A-Za-z]/);

        chomp;
        # Windows line endings, just in case...
        tr/\015//d;

        my ($date, $weight, @factors) = split($SepPat);

        next unless ((defined $date) and $date =~ /^\d/);

        # Only generate a training set if everything is defined and
        # we have a prior day weight to compare to
        unless ((defined $weight) and looks_like_number($weight)) {
            $weight = '' unless (defined $weight);
            if ($Interactive) {
                warn "$ARGV:$. weight: '$weight' is not a number - line ignored\n";
            }
            undef $prev_weight;
            next;
        }

        
        #
        # -- collect daily (gain + factors) data points in @daily
        #
        if ((defined $prev_weight) && scalar(@factors) > 0) {
            my $gain = $weight - $prev_weight;
            my @day_list = ($gain, @factors);
            push(@daily, \@day_list);
        }
        $prev_weight = $weight;
    }

    #
    # Output vw training-set
    #
    for (my $i = 0; $i < @daily; $i++) {
        my $start = $i;
        my $end = $start + $NDays - 1;
        my $sum_gain = 0.0;
        my @sum_factors = ();
        for (my $j = $i; $j <= $end; $j++) {
            #
            # -- Aggregate consecutive days gains and factors - up to $NDays
            #
            my $day_list = $daily[$i];
            my @gain_factors = @$day_list;
            my ($gain, @factors) = @gain_factors;
            $sum_gain += $gain;
            push(@sum_factors, @factors);
            printf "%.2f | @sum_factors\n", $sum_gain;
        }
    }
}

#
# -- main
#
process_args();
process_input();