Tweets Classification using Different Classifiers for Sentiment Analysis

Abstract

Twitter is a famous platform for social networking where people share and engage with "tweets" comments. It is a way of expressing people's opinions or emotions on various topics. Different parties, like producers and consumers, have carried out a sentiment analysis on tweets to gain insights into goods or to analyze the sector. Also, the accuracy of our sentiment analysis forecasts will increase with the recent advances in machine learning algorithms. In this research, we will try to analyze sentiments in tweets using different techniques of ML. We try to determine if it's positive or negative with the polarity of the tweet. The more powerful sentiment should be chosen as the final tweet if both positive and negative elements are included. The Kaggle dataset was crawled and positively/negatively labeled. The presented data include emoticons, identification, date, query, usernames, and information that have to be processed into a standard form. It is also essential to extract useful features from the text that is a type of "tweet" representation. We have often developed numerous classifications to classify sentiments using the extracted features. The objective of this work is to evaluate the results of various machine learning algorithms in the measurement of Twitter data. The efficiency by sentence classification of Decision Tree, Random Forest & XGB algorithms was compared. The findings show that by using Decision Tree & XGB, Random Forest has achieved great accuracy.