International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 06 Issue: 04 | Apr 2019 www.irjet.net p-ISSN: 2395-0072 © 2019, IRJET | Impact Factor value: 7.211 | ISO 9001:2008 Certified Journal | Page 4124 Real Time Sentiment Analysis of Political Twitter Data Using Machine Learning Approach Joylin Priya Pinto 1 , Vijaya Murari T. 2 1 Department of Computer Science and Engineering, NMAMIT, Nitte, Karnataka, India 2 Professor, Department of Computer Science and Engineering, NMAMIT, Nitte, Karnataka, India ---------------------------------------------------------------------***--------------------------------------------------------------------- Abstract - Social media is an interesting platform to directly measure people’s feelings. Communication technologies play a vital role in geographically locating these emotions. But, the understanding is not a simple task. Generation of voluminous data makes the manual process complicated. Social media data face the problem of diversity of language; due to which the automatic approach also becomes a tedious job. Social network deals with controversial discussions on variety of topics. These discussions help in the field of data analysis. In this paper, we are going to analyze tweets on political Ayodhya issue. Tweets of users are collected and analysis is done with the help of machine learning algorithm, to classify the polarity of tweets. Key Words: Sentiment Analysis, Feature Extraction, Support Vector Machine, Random Forest, Naïve Bayes, Linear Regression, KNN 1.INTRODUCTION Analyzing the sentiments of people is turning into a critical viewpoint in a wide range of decision making process since it is useful in recognizing individual’s issues and strategies strengths. Without a doubt, these information can be utilized to settle on increasingly educated choices which will probably conclude in better utilization of assets, good association, better administration, improved citizen lifestyle, nice human relations and, in the long run, better society. Taking are of large scale classification issues is very essential in numerous applications, for example, text classification problem. There is a high possibility of facing distinctive troubles while performing sentiment analysis; for each situation people may not present their assumptions similarly, a solitary sentence may be certain for one event and can be negative for other event; an enormous number of sentence mixes are possible. Identifying incorrect spelling, and dealing with intensifiers and fake sentences are very challenging. In the past few years, people feedback and sentiments were analyzed by opinion pooling system that means with the help of conducting interviews, questionnaires and by collecting opinions through forms, [1][2] now a days social media is the best way to analyze people sentiments [3]. There is an immense growth of user- generated data in the form of blogs, forum and tweets. With the increased usage of these platforms, people started to speak about all matters: from personal to public; general to specific matters. Social media is an effective platform to understand individual sentiments. The analysis of user posts can be utilized to take proper decisions in a variety of fields such as Business, Election, Product review, Government, and so on. Now a days, utilizing social media for political discussions has become a common practice. Political campaigns have misused immense range of information accessible on Twitter to draw insights about people sentiments and in this way structure their promoting efforts. Different sentiment analysis algorithms can be utilized to distinguish and break down the attitudes of the users towards a political talk. Furthermore, with the recent advancements in machine learning algorithms, it became possible to enhance the exactness of sentiment analysis predictions. An effective online social micro-blogging service is Twitter. It allows users to communicate with short messages of length 140-characters. These messages are called as "tweets". Twitter is specially an interesting platform because of the usage of its hash tags. Along with the short messages, users can use hash tag symbol ‘#’ before a specific keyword or phrase in the Tweet to arrange the tweets and make them easier in Twitter Search. The issue of text classification can be made relatively simpler since the hash tag itself can express a feeling or sentiment. In this work, twitter data on Ayodhya issue is used as data source for sentiment analysis. 1.1 Sentiment Classification Sentiment classification is one of the important topics in sentiment analysis, which classifies the expressed opinion towards an entity as positive, negative or unbiased. Three most essential categorical levels in Sentiment Analysis are Document level, Sentence level, and Feature based Analysis [4]. The underlying stage in Sentiment Analysis is to perceive whether the sentence is opinionated or objective. Subjective sentence conveys emotions whereas objective sentence does not convey any emotion; usually considered as unbiased and also being ignored in the analysis. Sentence level classification is different from that at document level [5]. A document can be more or less opinionated, whereas a sentence can be only subjective or objective. Comparing with the document level sentiment classification, sentence level classification has one more task to do. There is a need to