Sentiment-Analysis

Create an api to take user input statement, clean the dataset, model the data, and classify the statement as positive, negative or neutral.

Import libraries to process your data

import nltk
nltk.download('wordnet')
nltk.download('averaged_perceptron_tagger')
nltk.download('punkt')

Download twitter dataset for training your model

from nltk.corpus import twitter_samples

Libraries to clean your data

from nltk.tag import pos_tag
from nltk.stem.wordnet import WordNetLemmatizer
from nltk.corpus import stopwords
nltk.download('stopwords')

Classifier

from nltk import classify, NaiveBayesClassifier

Pickle file

We have used a pickle file to store our model, since it conumes a lot of time for cleaning and training our model.

import pickle as pickle
pickle.dump(classifier, open('models/final_prediction.pickle', 'wb'))

Chcek results

First, run the Jupyter notebook file. Second, open "test.py" file and run it in cmd. Server starts running and url is displayed, mostly - http://127.0.0.1:5000. Open postman - and enter url with POST method: http://127.0.0.1:5000/sentiment Request body: { "data": "I did not like the movie at all. It was so boring!!!" }

For this classifier, I've used 1500 positive tweets, 1500 negative tweets and 3000 neutral tweets for training. The larger your dataset, the more accurate your model is.