Twitter Sentiment Analysis using Machine Learning Algorithms for COVID-19 Outbreak in New Zealand

Abstract

The use of data obtained from social media for data mining has benefited the process of analysing public opinions. Sentiment analysis, also referred to as opinion mining, helps public health officials and other governmental agencies to understand the public's concerns, panics, emotions and interactions to provide effective services and information. This research focuses on the sentiment analysis of the COVID-19 pandemic outbreak in New Zealand using Twitter data. The analyses derived from several machine learning classification techniques, in particular Naive Bayes, K-Nearest Neighbour (KNN), Convolutional Neural Network (CNN), and Support Vector Machine (SVM). In addition, the researchers implemented these algorithms in two different data mining platforms, namely Python and RapidMiner, then compared the metrics obtained from these techniques and platforms to identify the best algorithm suited for sentiment analysis. Finally, the researchers illustrate the experimental results that show the performance of Naïve Bayes and SVM, which indicates a longer computational time and led to an improved Twitter sentiment analysis result that outperforms the other models. After that, validate these models' effectiveness by comparing the obtained results for the same models in Python and RapidMiner.