bin/single_valued_columns

   1 #!/usr/local/bin/perl
   2
   3 use FindBin qw($Bin);
   4 use lib "$Bin/../lib";
   5
   6 # Don't edit the line below, it must look exactly like this.
   7 # Everything above this line will be replaced #
   8
   9 use PsN;
  10 use data;
  11 use strict;
  12
  13 use Getopt::Long;
  14 use vars qw/ $opt_help /;
  15 use vars qw/ $opt_data
  16              $opt_idcolumn
  17              $opt_subset_name
  18              $opt_remainder_name
  19              $opt_do_not_test_columns
  20              $opt_debug
  21              $opt_debug_package
  22              $opt_debug_subroutine /;
  23
  24 ## Configure the command line parsing
  25 Getopt::Long::config("auto_abbrev");
  26
  27 ## Declare the options
  28 my $res = GetOptions("help",                 # Display help message
  29                      "idcolumn:i",
  30                      "data:s",
  31                      "subset_name:s",
  32                      "remainder_name:s",
  33                      "do_not_test_columns:s",
  34                      "directory:s",
  35                      "debug:0",
  36                      "debug_package:s",
  37                      "debug_subroutine:s" );
  38
  39 exit unless $res;
  40
  41 if($opt_help or not defined $opt_data) {
  42   print <<'ENDHELP';
  43
  44   single_valued_columns
  45
  46       Perl script for identification of columns which contain only one
  47       value per individual. Creates two new data files: one containing the
  48       columns that are single valued and one containing the rest.
  49
  50     Usage:
  51
  52       single_valued_columns -data=filename
  53       [-idcolumn=column_number]
  54       [-subset_name=single_value_filename]
  55       [-remainder_name=remaining_data_filename]
  56       [-do_not_test_columns='2,3,5..7']
  57
  58     Example:
  59
  60       perl single_valued_columns -data=pheno.dta -idc=1 -sub=mysub.dta -rem=myrem.dta -do_not='2,3'
  61
  62 ENDHELP
  63
  64   exit;
  65 }
  66
  67 ## Check that we do have a data file
  68 unless ( $opt_data ) {
  69   die "A data file must be specified\n";
  70   return;
  71 }
  72
  73 ui -> category( 'data' );
  74
  75 debug -> level( $opt_debug );
  76 debug -> package( $opt_debug_package );
  77 debug -> subroutine( $opt_debug_subroutine );
  78
  79 my @dont_test = split(',',$opt_do_not_test_columns);
  80
  81 my $data = data -> new ( filename => $opt_data,
  82                          idcolumn => $opt_idcolumn );
  83
  84 $data -> single_valued_data ( subset_name         => $opt_subset_name,
  85                               remainder_name      => $opt_remainder_name,
  86                               do_not_test_columns => \@dont_test,
  87                               target              => 'disk' );