COVID-19 is acknowledged as a transmitted from one person to another through contact, coughing, and sneezing. Twitter has served as one of the media outlets to raise awareness regarding COVID-19 problems. One of the government's objectives, based on the rising distribution, is pursued to preserve immunizations in stock. Hence, the vaccine information has become adequately available. However, immunization has sparked a range of reactions, including support and objection for vaccination. Attempts require a mechanism to distinguish tweets addressing immunization-related information. One notable method includes sentiment analysis, expressing a statement's negative, neutral, and positive feelings. A total of 5200 datasets were employed, with 4000 datasets classified as neutral, 300 datasets as negative, and 900 datasets as positive. The Naïve Bayes method and the TF-IDF (Term Frequency – Inverse Document Frequency) word weighting strategy are proposed to model the COVID-19 vaccine dataset, by comparing the three models of: Gaussian, Multinomial, and TF-IDF (Term Frequency – Inverse Document Frequency). According to study employing Naïve Bayes, the best model employing Bernoulli Naive Bayes is 80% with a data splitting of 30%.