Abstrait

Election Prediction Using Twitter Sentiment Analysis

Rashiduzzaman Prodhani, Atowar Ul Islam and Luit Das

Sentiment analysis is the computational study of opinions, sentiments, evaluations, attitudes, views and emotions expressed in text. It refers to a classification problem where the main focus is to predict the polarity of words and then classify them into positive or negative sentiment. Sentiment analysis over Twitter offers people a fast and effective way to measure the public’s feelings towards their party and politicians. The primary issues in previous sentiment analysis techniques are classification accuracy, as they incorrectly classify most of the tweets with the biasing towards the training data. So I collected live data to predict the accurate election result. Twitter is a place where users posting quick and real-time updates about different activities or events as the spread of information and news are quick enough. We used the python library “Tweepy” for accessing the Twitter API and fetched live data from Twitter. More than 2000 tweets for each political party candidate are fetched by using keywords. Using “TextBlob” library of python, sentiments are applied to each tweet and depending upon more positive tweets for particular candidate and we can visualize a prediction. Text classification algorithms like Naive Bayes, Support Vector Machine (SVM) and Random Forest are used to train model using labelled data. The accuracy of the predicted result is calculated and the result is declared finally, result is represented in the form of pie chart, bar graph for each political candidate representing positive, negative and neutral sentiments.

Social media; Twitter; Politics; Sentiment analysis; Naive bayes; Support vector machine; Positive; Negative; Neutral