Python Code to train a Hidden Markov Model, using NLTK

hmm-example.py

__author__ = 'ssbushi'

# Import the toolkit and tags
import nltk
from nltk.corpus import treebank

# Train data - pretagged
train_data = treebank.tagged_sents()[:3000]
test_data = treebank.tagged_sents()[3000:]

print(train_data[0])

# Import HMM module
from nltk.tag import hmm

# Setup the tagger
tagger = nltk.HiddenMarkovModelTagger.train(train_data)

print(tagger)
# Prints the basic data about the tagger

print(tagger.tag("Today is a good day .".split()))

print(tagger.evaluate(test_data))

"""
Output in order (Notice some tags are wrong :/):
[('Today', u'NN'), ('is', u'VBZ'), ('a', u'DT'), ('good', u'JJ'), ('day', u'NN'), ('.', u'.')]
0.36844377293330455
"""