<![CDATA[Bank SeaBank is a digital bank that has social media accounts on Twitter (X) and YouTube with thousands of followers. These platforms are often used to express opinions or comments on various topics. This research aims to provide a benchmark for Bank SeaBank to improve its services based on positive and negative user reviews. The data analyzed consists of 500 comments on Twitter (X) and YouTube with the keyword "Bank SeaBank." The methods used include machine learning algorithms such as SVM, Naïve Bayes, k-NN, Decision Tree, Logistic Regression, as well as the deep learning algorithm pre-trained BERT. The analysis results show the highest accuracy for SVM at 84%, followed by Naïve Bayes at 81%, k-NN at 80%, and both Decision Tree and Logistic Regression at 77%. The deep learning algorithm BERT achieved an accuracy of 86% with 3 epochs and a training-to-testing data ratio of 80:20.Kata kunci: SeaBank; Social media; BERT algorithm AbstrakBank SeaBank adalah salah satu bank digital yang memiliki media sosial Twitter (X) dan YouTube dengan ribuan pengikut. Kedua platform ini sering digunakan untuk menyampaikan pendapat atau komentar tentang berbagai topik. Penelitian ini bertujuan untuk memberikan tolak ukur bagi Bank SeaBank dalam meningkatkan layanan berdasarkan ulasan positif dan negatif dari pengguna. Data yang dianalisis terdiri dari 500 komentar di Twitter (X) dan YouTube dengan kata kunci "Bank SeaBank". Metode yang digunakan mencakup algoritma machine learning seperti SVM, Naïve Bayes, k-NN, Decision Tree, Logistic Regression, serta algoritma deep learning pre-trained BERT. Hasil analisis menunjukkan akurasi tertinggi pada SVM sebesar 84%, diikuti oleh Naïve Bayes sebesar 81%, k-NN sebesar 80%, Decision Tree dan Logistic Regression masing-masing sebesar 77%. Algoritma deep learning BERT mencapai akurasi 86% dengan 3 epoch dan proporsi data latih dan uji sebesar 80:20Kata Kunci: SeaBank; Media Sosial; Algoritma BERT]]>
Copyrights © 2024
