HABIBA, Azifa and Isnanto, Rizal and Suseno, Jatmiko Endro (2022) PENGARUH SELEKSI CIRI CHI SQUARE PADA ALGORITMA NAIVE BAYES TERHADAP ANALISIS SENTIMEN MASYARAKAT INDONESIA TENTANG PEMBELAJARAN TATAP MUKA DI MASA PANDEMI COVID-19. Masters thesis, School of Postgraduate Studies.

	Text 1) Cover.pdf Download (1MB)
	Text 2) BAB I.pdf Download (97kB)
	Text 3) BAB II.pdf Download (447kB)
	Text 4) BAB III.pdf Restricted to Repository staff only Download (368kB)
	Text 5) BAB IV.pdf Restricted to Repository staff only Download (824kB)
	Text 6) BAB V.pdf Download (91kB)
	Text 7) Daftar Pustaka.pdf Download (216kB)

Abstract

Kritik dan komentar yang disampaikan masyarakat Indonesia terkait kebijakan pemerintah mengenai pembelajaran tatap muka di masa pandemi Covid-19 menuai pro dan kontra. Tidak sedikit orang tua yang khawatir dengan kebijakan ini dikarenakan ketakutan akan penyebaran klaster baru Covid-19 di Indonesia yang semakin berkembang, di sisi lain pembelajaran secara daring juga dinilai kurang efektif karena banyak siswa yang sulit menerima materi yang disampaikan guru serta banyaknya siswa yang belum memiliki perangkat yang memadai. Opini yang dituliskan pada Twitter membutuhkan pengklasifikasian sesuai sentimen yang dimiliki agar mudah untuk mendapatkan kecenderungan opini tersebut apakah cenderung beropini netral, positif maupun negatif. Analisis sentimen dalam penelitian ini dilakukan dengan menggunakan metode Naïve Bayes dan seleksi ciri Chi Square dalam melakukan klasifikasi. Dari hasil penelitian yang telah dilakukan didapatkan bahwa sentimen masyarakat tentang pembelajaran tatap muka di tengah pandemi Covid-19 mendapatkan respon yang cenderung positif sebesar 65,78% dan sentimen negatif sebesar 34,22%. Masyarakat lebih banyak memberikan sentimen positif terkait pembelajaran tatap muka di masa pandemi, sebagian dari sentimen positif berisi doa dan harapan dalam menyambut pembelajaran tatap muka. Namun tidak sedikit juga masyarakat yang memberikan sentimen negatif, beberapa sentimen negatif berisi kekecewaan yang mana salah satu penyebabnya adalah adanya wacana terkait pembelajaran tatap muka di daerah tempat tinggalnya tetapi belum terjadi. Hasil analisis dari metode Naïve Bayes dengan seleksi ciri chi Square memiliki akurasi 93% dan tanpa seleksi ciri Chi Square memiliki akurasi 92%, sehingga dapat disimpulkan Metode Naïve Bayes dengan seleksi ciri Chi Square memiliki tingkat akurasi yang lebih baik dibanding tanpa menggunakan seleksi ciri Chi Square
Kata Kunci: Analisis Sentimen, Chi Square, Covid-19, Naïve Bayes, Seleksi Fitur

Criticisms and comments submitted by the Indonesian people regarding government policies related to face-to-face learning during the Covid-19 pandemic reap the pros and cons. Not a few parents are worried about this policy because they are worried about the increasingly widespread spread of the new Covid-19 cluster in Indonesia, on the other hand online learning is also considered less effective because many students have difficulty accepting the material presented by the school. teachers and many students do not have adequate tools. Opinions written on Twitter require classification according to the sentiment they have so that it is easy to get the tendency of opinions whether they tend to be neutral, positive or negative. Sentiment analysis in this study was carried out using the Naïve Bayes method and Chi Square feature selection in its classification. From the results of the research that has been carried out, it was found that public sentiment towards face-to-face learning in the midst of the Covid-19 pandemic received a positive response of 65.78% and negative sentiment of 34.22%. The positive sentiment of the community is more related to face-to-face learning during the pandemic, some of these positive sentiments contain prayers and hopes in welcoming face-to-face learning. However, not a few also gave negative sentiments, some of these negative sentiments contained disappointment, one of which was the discourse related to face-to-face learning in the area where they lived but it had not yet happened. The results of the analysis of the Nave Bayes method with chi Square feature selection has an accuracy of 93% and without Chi Square feature selection has an accuracy of 92%, so it can be concluded that the Naïve Bayes method with Chi Square feature selection has a better accuracy rate than without using Chi Square characteristic selection
Keywords: Sentiment Analysis, Chi Square, Covid-19, Naïve Bayes, Feature Selection

Actions (login required)

View Item