nltk-classifier-helpers

Install in a virtualenv

virtualenv virtualenv
. virtualenv/bin/activate
pip install PyYAML # need to install separately, before nltk
pip install nltk==2.0.4 # did not test on 3.0
pip install numpy # needed for the MaxentClassifier
pip install -e [email protected]:takdavid/nltk-classifier-helpers.git#egg=nltk-classifier-helpers 

Train classifiers on your test input file

python learn.py testdata.txt

This will store the classifier objects in pkl files, as well as ready-to-use python code for 1..6 levels deep decision tree classifiers.

The format of the input datafile:

• frist column: id
• last column: expected label
• all the others are features:
• feature without colon: binary feature
• feature with colon: key-value pair

Evaluate your classifiers in pkl files on some test data

python evaluate.py testdata.NaiveBayesClassifier.pkl testdata.txt
python evaluate.py testdata.DecisionTreeClassifier.pkl testdata.txt
python evaluate.py testdata.MaxentClassifier.pkl testdata.txt

Use your classifier to classify some test data

python classify.py testdata.NaiveBayesClassifier.pkl testdata.txt