Komparasi Model Analisis Sentimen Pada Twitter Terhadap Kemahalan Minyak Goreng dengan Metode Naive Bayes dan Support Vector Machine

Abstract

At the end of 2021, people are shocked by the drastically reduced supply of cooking oil and high prices. This makes people talk about it a lot through social media like Twitter. Freedom on Twitter raises many responses from the public. The number of negative and positive responses on Twitter makes comparisons between the two responses difficult to observe. This study aims to determine the comparison of positive responses and negative responses. Machine learning with the naïve Bayes method and support vector machine is able to overcome this problem. The research conducted examines how the comparison between positive responses and negative responses and which method has higher accuracy. The data used is 10,000 Indonesian language tweets. Model testing was carried out with 1839 test data. the Naive Bayes method gets an accuracy of 74.06% with the results of predicting two positive tweets and 1837 negative tweets. The SVM method was tested on linear, polynomial, RBF, and sigmoid kernels. The kernel with the highest accuracy value is the sigmoid kernel with an accuracy of 81.8% with the predicted results of 266 positive tweets and 1573 negative tweets.