Sentiments Analysis of COVID-19 Subvariant XBB.1.5 using machine learning

Abstract

Abstract The Covid-19 pandemic has caused a significant impact on society, with discussions about the virus taking place on various social media platforms. In this study, different machine learning techniques for sentiment analysis of COVID-19 Subvariant XBB.1.5 were explored. The datasets of tweets containing hashtags related to Covid-19 Subvariant XBB.1.5 were collected and natural language processing techniques were used as processing techniques to pre-process the text. In this research, all tweets related to COVID-19 Subvariant XBB.1.5 from October 15th, 2022 are collected using the Twitter API. Different machine learning algorithms were later used to classify the tweets as positive, neutral, or negative in sentiment. The different algorithms used includes Stochastic Gradient Descent, Logistic regression, Naïve Bayes, Random Forest, Support Vector Machine and Extreme Gradient Boosting Our results showed that Logistic Regression achieved the highest accuracy, with an overall accuracy of 89% testing accuracy and 100% training accuracy, with positive sentiments having 0.95 Precision, 0.9 recall, and 0.93 F1-Score. The higher percentage of Positive tweets sentiments about COVID-19 Subvariant XBB.1.5, showed that most people were not disturbed about the negative impact the virus could have on them in comparison to the first and other previous Covid variants. The statistical performance of the different machine learning algorithms is measured using Accuracy, F1-Score, recall, precision, and ROC. The accuracy of the different classifiers applied is high.