classifierTrain.py

# -*- coding: utf-8 -*-
"""Sentiment Analysis Demo VIP.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1bD3Zr3yNVoVYtd7vIhCzkQKTM39G-ZYf
"""

import pandas as pd
import numpy as np

from sklearn.feature_extraction.text import CountVectorizer
from sklearn.naive_bayes import BernoulliNB

#Upload files to this notebook
# from google.colab import files
# uploaded = files.upload()

#Combine the files

with open("classifierData/imdb_labelled.txt", "r") as text_file:
    lines = text_file.read().split("\n") 

with open("classifierData/amazon_cells_labelled.txt", "r") as text_file:
    lines = text_file.read().split("\n") 

with open("classifierData/yelp_labelled.txt", "r") as text_file:
    lines = text_file.read().split("\n")

newLines = [line.split("\t") for line in lines if len(line.split("t")) == 2 and line.split("\t")[1] != ""]

#Split data into Train Features & Train Labels:
train_documents = [line[0] for line in newLines]

train_labels = [int(line[1]) for line in newLines]

#Convert the training set to a matrix of token counts:

count_vectorizer = CountVectorizer(binary="true")
train_documents = count_vectorizer.fit_transform(train_documents)

#Fit the BernoulliNB Classifier:

classifier = BernoulliNB().fit(train_documents, train_labels)

def predictionOutput(sentence):
  """
  This function outputs the sentiment analysis label (Positive or Negative) for a given sentence
  """
  prediction = classifier.predict(count_vectorizer.transform([sentence]))
  if(prediction[0] == 1):
      print("This is a Positive Sentiment")
  elif (prediction[0] == 0):
      print("This is a Negative Sentiment")

#Testing

predictionOutput("I am having a very good and great day")

predictionOutput("I hate this terrible movie")