Analisis Sentimen Komentar Youtube Tentang Relawan Patwal Ambulance Menggunakan Algoritma Naïve Bayes dan Decision Tree

Abstract

The presence of the ambulance patrol is sometimes considered disturbing by the community, many people's opinions are still pro and contra about the actions carried out by civilian ambulance patrol volunteers because they are considered illegal and sometimes also arrogant. In this study. the researchers wanted to know the opinions and responses of the public about the actions taken by civilian ambulance patrol volunteers. The method used in this study is to perform sentiment analysis with data mining techniques to find the polarity of sentences in a document using the Naïve Bayes and Decision Tree algorithms. The initial steps in this research were collecting comment data by scraping YouTube comments using API Key Youtube V3, followed by manual data labeling, data cleaning, data preprocessing, and word weighting using TF-IDF. From the overall results of testing with 600 training data, the Naïve Bayes algorithm has a higher accuracy value of 66.72% while the Recall value is 64.98%. Testing with the Decision Tree algorithm in this study has a higher Recall compared to Naïve Bayes. %. From the results of the YouTube comment dataset used in this study, it can be concluded that the Naïve Bayes algorithm has a higher accuracy value than the Decision Tree algorithm, so it can be concluded that Naïve Bayes has the best accuracy in the YouTube comment dataset used in this study.