Abstract
The world went through an anomalous time, showcasing COVID-19 on the top. Staying home is the best solution to be safe. People spend most of their time on social media platforms. They are busy in reacting on public posts, news, articles, and commenting there. Now a days, even the media organizations and government agencies are providing the latest news and opinions related to coronavirus, so every person's comment and post speak high about his sentiments. Therefore, the need to analyze text sentiment has arisen. It is helpful to get an opinion about a specific subject effectively. In this paper, we aim to understand and analyze the quarantine impact of coronavirus pandemic (COVID-19) on our lives. In fact, we used multiclass Sentiment Analysis (SA) to classify people's opinions from extracted tweets about COVID-19 quarantine impact, using first different ma- chine learning (ML) classifiers such as Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Multinomial Naive Bayes (MNB) and secondly various Deep Learning (DL) models such as Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM). In our experimental result, for the ML models, the result of SA showed that SVM gives the best performance with an accuracy of 83.11% for DL models, RNN- LSTM achieved an accuracy 86.53% and AUC of about 93.21%. Practical experiments are performed on 8111 tweets dataset, reporting positive performances, which are competitive to other well-known approaches.