search for: primary_terms

...---------------------------# # Perl Search Script #------------------------------------------------------------------------------# #!/usr/bin/perl -w use strict; use ExtUtils::testlib; use Search::Xapian qw/:all/; #------------------------------------------------------------------------------# my $primary_terms = "Swebmd.com Shealth.com Smyhealthcare.com Shealthfinder.gov"; my $secondary_terms = "fitness Diseases diet science "; $secondary_terms =~ tr/A-Z/a-z/; # convert to lower case #------------------------------------------------------------------------------# my $db = Search::Xa...

...nd 4 others like it but which did not make it to final model anyway), and then running stepAIC. Question: Why do these processes not yield same model? Here are all the details if helpful: I start with 12 potential predictor variables, 7 "primary" terms and 5 additional that are I(primary_terms^2). I run a glm for these 12 and then do stepAIC (BIC actually) both directions. The scope argument is scope=list(upper=~.^2,lower=NULL). This means there are 78 predictor terms considered, the 12 primary terms and 66 interactions [n(n+1)/2]. I see this with trace=T also. Here is the code used...