A Covid-19 Sentiment Analysis on Twitter Using K-Nearest Neighbours

Abstract

In December 2019, an outbreak named Corona Virus (SARS-CoV-2) occurred in the city of Wuhan, China which was later known as COVID-19. News of the development of the virus spread through various media, one of which was through the well-known platform Twitter. Twitter is one of the widely used media platforms to communicate about Covid-19. Information related to Covid-19 circulating in the community can be in the form of news or opinions or opinions. Then, the circulating information will be classified into three classes, namely positive, negative or neutral. The method used to calculate the prediction of text classification on Twitter is K-nearest neighbors (KNN). The dataset used in grouping on twitter by using the account name Covid19. Firstly, the dataset by crawling data or information on twitter. Secondly, the text mining stage to determine the class distance value and calculate the Euclidean distance formula based on all the training data to be tested. After the training process is complete, the evaluation model used will be used, the Euclidean results are taken based on the value of the closest distance. The accuracy of the model will be calculated using the previous Euclidean method. The results of this study he obtained with the highest value, one of which was 78% using a 50:50 sample comparison with k-5 and k-9 values.