Home > chi2

chi2

Chi2 is a project mainly written in C, it's free.

This is a C implementation of the chi2 feature selection used on information retrieval.

Chi2

A pure-C implementation of the chi2 feature selection that can handle about hierarchy, quite used on text classification.

Compiling

This program should compile on any gcc version, but it was only tested on Linux and Mac OS X. A Makefile is included, so to compile, just go into the chi2 directory and:

$ make

Running

The program can be executed as:

    $ chi2 -i [input file] -o [output file] -n [number of features to be selected] -c [base class] -s [class separator]

The input file format is an arff file, used by Weka.

There is a test file on 'test/0.arff ', and the 'make run' option should run this test against "comp.sys" selecting 80 features.

Roadmap

Plans for any (but unlikely) future updates are the inclusion of other techniques (Information Gain) and support of other file formats (XML, CSV).

TODO

update all comments on code to English

Previous:moja-stronka