perplexity score nltk

[Shengrong Liu]

Dear All,

I did not use nltk package for unigram model, and I had nearly 100 lines. It will be a lot complicated for bigram or trigram. I heard that NLTK can calculate perplexity score. 

This is so far I have 

import nltk

from nltk.model.ngram import NgramModel

from nltk.probability import LidstoneProbDist

f_in= open(r'C:\Python27\A.txt')

ln = f_in.read()    

words = nltk.word_tokenize(ln)

my_bigrams = nltk.bigrams(words)

my_trigrams = nltk.trigrams(words)

estimator = lambda fdist, bins: LidstoneProbDist(fdist, 0.2)

what will be the next？ 

what is the estimator？ 

Is there any step I need to do before use .perplexity()

Thanks