Text::NSP::Measures::2D::Fisher2 - Perl module that provides methods to compute the Fishers exact tests.
use Text::NSP::Measures::2D::Fisher2::left; my $npp = 60; my $n1p = 20; my $np1 = 20; my $n11 = 10; $left_value = calculateStatistic( n11=>$n11, n1p=>$n1p, np1=>$np1, npp=>$npp); if( ($errorCode = getErrorCode())) { print STDERR $errorCode." - ".getErrorMessage(); } else { print getStatisticName."value for bigram is ".$left_value; }
This module provides a framework for the naive implementation of the fishers exact tests. That is the implementation does not have any optimizations for performance. This will compute the factorials for the hypergeometric probabilities using direct multiplications.
This measure should be used if you need exact values without any rounding errors, and you are not worried about the performance of the measure, otherwise use the implementations under the Text::NSP::Measures::2D::Fisher module.
To use this implementation, you will have to specify the entire module name. Usage:
statistic.pl Text::NSP::Measures::Fisher2::left dest.txt source.cnt
Assume that the frequency count data associated with a bigram <word1><word2> is stored in a 2x2 contingency table:
word2 ~word2 word1 n11 n12 | n1p ~word1 n21 n22 | n2p -------------- np1 np2 npp
where n11 is the number of times <word1><word2> occur together, and n12 is the number of times <word1> occurs with some word other than word2, and n1p is the number of times in total that word1 occurs as the first word in a bigram.
The fishers exact tests are calculated by fixing the marginal totals and computing the hypergeometric probabilities for all the possible contingency tables,
A left sided test is calculated by adding the probabilities of all the possible two by two contingency tables formed by fixing the marginal totals and changing the value of n11 to less than the given value. A left sided Fisher's Exact Test tells us how likely it is to randomly sample a table where n11 is less than observed. In other words, it tells us how likely it is to sample an observation where the two words are less dependent than currently observed.
A right sided test is calculated by adding the probabilities of all the possible two by two contingency tables formed by fixing the marginal totals and changing the value of n11 to greater than or equal to the given value. A right sided Fisher's Exact Test tells us how likely it is to randomly sample a table where n11 is greater than observed. In other words, it tells us how likely it is to sample an observation where the two words are more dependent than currently observed.
A two-tailed fishers test is calculated by adding the probabilities of all the contingency tables with probabilities less than the probability of the observed table. The two-tailed fishers test tells us how likely it would be to observe an contingency table which is less probable than the current table.
INPUT PARAMS : $count_values .. Reference of an array containing the count values computed by the count.pl program.
RETURN VALUES : 1/undef ..returns '1' to indicate success and an undefined(NULL) value to indicate failure.
INPUT PARAMS : $n11_start .. the value for the cell 1,1 in the first contingency table $final_limit .. the value of cell 1,1 in the last contingency table for which we have to compute the probability.
RETURN VALUES : $probability .. Reference to a hash containing hypergeometric probabilities for all the possible contingency tables
Ted Pedersen, University of Minnesota Duluth <tpederse@d.umn.edu>
Satanjeev Banerjee, Carnegie Mellon University <satanjeev@cmu.edu>
Amruta Purandare, University of Pittsburgh <amruta@cs.pitt.edu>
Bridget Thomson-McInnes, University of Minnesota Twin Cities <bthompson@d.umn.edu>
Saiyam Kohli, University of Minnesota Duluth <kohli003@d.umn.edu>
Last updated: $Id: Fisher2.pm,v 1.11 2008/03/26 17:18:26 tpederse Exp $
http://groups.yahoo.com/group/ngram/
http://www.d.umn.edu/~tpederse/nsp.html
Copyright (C) 2000-2006, Ted Pedersen, Satanjeev Banerjee, Amruta Purandare, Bridget Thomson-McInnes and Saiyam Kohli
This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.
This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with this program; if not, write to
The Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
Note: a copy of the GNU General Public License is available on the web at http://www.gnu.org/licenses/gpl.txt and is included in this distribution as GPL.txt.