Pemodelan Analisis Sentimen Masyarakat terhadap Adaptasi Kebiasaan Baru (AKB) mengunakan Algoritma Naïve Bayes

Abstract

AbstrakPandemi Covid-19 hampir masuk tahun ke dua di Indonesia, pemerintah terus berupaya menekan laju peningkatan penularan Covid-19 melalui berbagai media. Sosialisasi dan informasi melalui media sosial yang merupakan wadah paling cepat untuk tersampaikan kepada masyarakat. Berbagai istilah digunakan seperti adaptasi kebiasaan baru, social distancing, PSBB sampai PPKM sehingga memicu masyarakat untuk beropini di media sosial. Penelitian ini menganalisis sentiment masyarakat terkait opini peningkatan Covid-19 dari twitter. Klasifikasi tweet menggunakan Naive Bayes dengan penambahan seleksi fitur. Penggunaan confusion matriks untuk mengetahui performance algoritma Naive Bayes. Berdasarkan pengujian, penelitian ini menghasilkan 76% dengan accuracy positif sebesar 72,727%, accuracy negatif sebesar 75% dan accuracy netral sebesar 78,947%. Sehingga disimpulkan penggunaan model klasifikasi Naive Bayes dengan fitur seleksi dapat meningkatkan akurasi.Kata kunci: analisis sentimen, seleksi fitur, twitter crawling, naïve bayes, klasifikasi, emosiAbstractCovid-19 pandemic is almost in its second year in Indonesia, the government continues to try to suppress the rate of increase in the transmission of Covid-19 through various media. Socialization and information through social media which is the fastest medium to be conveyed to the public. Various terms are used, such as adapting new habits, social distancing, PSBB to PPKM, thus triggering the public to share opinions on social media. This study analyzes public sentiment regarding the increasing opinion of Covid-19 from twitter. Tweet classification based on positive, negative and neutral classes using Naive Bayes with feature selection. The use of confusion matrix to determine the performance of the Naive Bayes algorithm. BasedThis Research, the results from the sentiment analysis system using the nave Bayes classifier of 76% with positive accuracy of 72.727%, negative accuracy of 75% and neutral accuracy of 78.947%. So it can be concluded that the use of the Naive Bayes classification model with the selection feature can increase accuracy.Keywords: sentiment analysis, fitur selection twitter crawling, naïve bayes, clasification, emotion