203 SENTIMENT ANALYSIS OF TWITTER DATA ON DISTANCE LEARNING USING NAÏVE BAYES ALGORITHM Putri Rana Khairina 1 , Desti Fitriati 2 Informatics Engineering Study Program 1,2 Pancasila University 1,2 putrirnkh@gmail.com 1 , desti.fitriati@univpancasila.ac.id 2 Abstrak Covid-19 menyebar secara luas hingga mengakibatkan pandemi global. Sistem Pembelajaran Jarak Jauh (PJJ) dianggap sebagai solusi tetapi, kenyataan saat pelaksanaan PJJ ini belum sesuai dengan harapan dari masyarakat. Pengguna twitter banyak menuliskan pendapatnya terhadap PJJ. Kecenderungan sentimen masyarakat dapat digunakan sebagai salah satu cara membenahi sistem pendidikan yang ada di Indonesia dan dapat menjadi masukan bagi pemerintah dalam menyempurnakan metode PJJ yang sedang dilaksanakan. Maka, penelitian ini menghasilkan sebuah sistem yang dapat menganalisis sentimen tweet terhadap PJJ. Tweet tersebut didapat menggunakan Twitter API. Metode yang digunakan adalah Naïve Bayes untuk proses klasifikasi sentimen positif, negatif dan netral dengan menggunakan 600 data. Kemudian, dilakukan pembagian data 80% data training dan 20% data testing yang akan di text preprocessing terlebih dahulu. Tingkat akurasi analisis sentimen terhadap PJJ dengan metode Naïve Bayes menggunakan 3-fold Cross Validation menghasilkan rata – rata sebesar 93%. Kata kunci: Covid-19, Pembelajaran Jarak Jauh (PJJ), Naïve Bayes, Analisis Sentimen, k-Fold Cross Validation. Abstract Covid-19 is widespread, resulting in a global pandemic. Distance Learning System (DLS) is considered as a solution but, the reality of the implementation of DLS is not in accordance with the expectations of the community. Many twitter users wrote their opinions on DLS. The tendency of public sentiment can be used as a way to improve the existing education system in Indonesia and can be an input for the government in improving the DLS method that is being implemented. Thus, this study produced a system that can analyze tweet sentiment towards DLS. The tweet was obtained using the Twitter API. The method used is Naïve Bayes for the process of classification of positive, negative and neutral sentiments using 600 data. Then, data sharing is done 80% data training and 20% data testing that will be in the text preprocessing first. The accuracy of sentiment analysis of DLS using Naïve Bayes method using 3-fold Cross Validation produces an average of 93%. Keywords: Covid-19, Distance Learning System (DLS), Naïve Bayes, Sentiment Analysis, k-Fold Cross Validation INTRODUCTION Covid-19 is a new type of corona virus found in Wuhan, Hubei, China in 2019, named Coronavirus disease-2019 which is shortened to Covid-19. Covid-19 has since been found to have spread widely, resulting in a global pandemic. Corona Virus is a type of virus that can cause mild symptomatic illness to severe symptomatic illness and the worst impact is death(Siagian, 2020). This virus is very quickly transmitted to almost all countries, including Indonesia. The effect created by the Covid-19 pandemic has assaulted the wellbeing area, yet additionally assaulted different areas, for example, the travel industry, account, business, to transportation (Salam, 2020). At this time all actvities are carried out at home, with the aim of tackling the increasing number of Covid-19 sufferers in Indonesia. At present, the spread of Covid-19 can occur through air, surfaces contaminated by viruses, droplets and through human waste (CAKTI INDRA GUNAWAN & YULITA, 2020). One form of anticipation for Covid-19 countermeasures, the Government issued apolicy Social Distancing and Physical Distancing. One form of government countermeasures, namely Large Scale Social Restrictions (LSSR) for all activities that meet directly with other people, one of which is learning activities. Therefore, the education sector implements Distance Learning System (DLS) or online learning which is carried out at home. The Ministry of Education and Culture as the provider of national education issued a circular