Abstract:
Omicron is a new variant of the Corona virus that was first reported to the World Health
Organization (WHO) from South Africa on November 24, 2021. The amount of information
related to the Omicron virus that is spread in the form of text on the internet and on social
media sites like Twitter is of interest to researchers. This study aims to determine public
sentiment from Twitter users about the omicron virus, then determine sentiment classification
using the Nave Bayes Classifier and calculate the level of accuracy of the method in sentiment
analysis. The data used is 1350 data points using two data testing scenarios, namely the first
scenario, Split Validation, using a data ratio of 70%:30%, 80%:20%, and 90%:10%. The
second scenario uses cross-validation, divided into 10 tests. Sentiment classifiers are divided
into three classes: neutral, fearful, and fearless. The classification process is carried out by
means of data pre-processing and weighting using the TF-IDF technique. The final results
of the classification process get the highest accuracy values of 85%, 81% precision, and 86%
recall in the split validation test with a 90:10 data division. Testing data with cross-validation
resulted in the highest accuracy of 93,1% accuracy, 36,6% precision, and 98% recall, with
a fold value of 1.
