Amazon reviews classification

import nltk
import random

from nltk.sentiment.util import *
from pymongo import MongoClient

random.seed(2213)

# SENTIMENT ANALYSIS - AMAZON REVIEWS CLASSIFICATION

# Retrieves document (review) data from mongodb
# Returns: [ (list of words review1, cat review1), ...]
def getReviewData():
    client = MongoClient('localhost', 27018)
    client.test.authenticate('studenti', '_____', mechanism='SCRAM-SHA-1')

    ...

    return reviews

# Create train and test sets
reviews = getReviewData()
random.shuffle(reviews)
sentimAnalyzer = nltk.sentiment.SentimentAnalyzer()
allWordsNeg = sentimAnalyzer.all_words([mark_negation(doc) for doc in reviews])
unigramFeats = sentimAnalyzer.unigram_word_feats(allWordsNeg, min_freq=4)
sentimAnalyzer.add_feat_extractor(extract_unigram_feats, unigrams=unigramFeats)
featureSets = sentimAnalyzer.apply_features(reviews)
trainSet, testSet = ...

# Train and classify
classifier = sentimAnalyzer.train(nltk.classify.NaiveBayesClassifier.train, trainSet)
for key,value in sorted(sentimAnalyzer.evaluate(testSet).items()):
    print('{0}: {1}'.format(key, value))