Peningkatan Akurasi Analisis Sentimen Ulasan Pengguna Aplikasi Akulaku melalui Perbandingan SVM dan Naïve Bayes dengan Seleksi Fitur Chi-Square
DOI:
https://doi.org/10.61722/jssr.v4i5.12271Keywords:
Sentiment Analysis, Support Vector Machine, Naïve Bayes, Chi-Square, AkulakuAbstract
The rapid growth of financial technology (fintech) services in Indonesia has increased the adoption of digital finance applications such as Akulaku, which provides Buy Now Pay Later (BNPL), digital payment, and online lending services. Despite its popularity, Akulaku faces persistent service issues, including the suspension of BNPL services by the Financial Services Authority (OJK), loan-limit restrictions, and a lack of fee transparency, which are reflected in negative reviews on the Google Play Store. Given the large volume of reviews, manual analysis is impractical, requiring an automated machine-learning approach to identify user sentiment efficiently. This study examines the application of two classification algorithms, Support Vector Machine (SVM) and Naïve Bayes (NB), supported by Term Frequency-Inverse Document Frequency (TF-IDF) feature extraction, Chi-Square feature selection, and class-imbalance handling. The Sample, Explore, Modify, Model, Assess (SEMMA) method was used as the research framework. A total of 2,521 reviews were collected through web scraping, which, after cleaning and lexicon-based labeling, resulted in 1,757 valid reviews distributed into positive, neutral, and negative classes. Both models were built using TF-IDF and Chi-Square feature selection and evaluated across three data-splitting scenarios (80:20, 75:25, and 90:10) using accuracy, precision, recall, and F1-score. The results show that SVM combined with Chi-Square feature selection consistently outperformed Naïve Bayes across all tested scenarios, achieving the best accuracy of 86.93% compared to 76.14% for Naïve Bayes. The best SVM model was further confirmed through 10-fold cross-validation, yielding a stable mean macro F1-score of 0.8033. The review analysis revealed that the main user complaints centered on loan-limit issues, loan-application rejection, and application technical stability, providing practical insights for service improvement.
Keywords: Chi-Square; Fintech; Naïve Bayes; Sentiment Analysis; Support Vector Machine
References
Cahyani, P., & Abdillah, L. (2024). Perbandingan Performa Algoritma Naïve Bayes , SVM dan Random Forest : Studi Kasus Analisis Sentimen Pengguna Sosial Media X. 11(02), 12–21.
Chairunnisa, C., Ernawati, I., & Santoni, M. M. (2022). Klasifikasi Sentimen Ulasan Pengguna Aplikasi PeduliLindungi di Google Play Menggunakan Algoritma Support Vector Machine dengan Seleksi Fitur Chi-Square. Informatik : Jurnal Ilmu Komputer, 18(1), 69. https://doi.org/10.52958/iftk.v17i4.4594
Ernayanti, T., Mustafid, M., Rusgiyono, A., & Hakim, A. R. (2023). Penggunaan Seleksi Fitur Chi-Square Dan Algoritma Multinomial Naïve Bayes Untuk Analisis Sentimen Pelangggan Tokopedia. Jurnal Gaussian, 11(4), 562–571. https://doi.org/10.14710/j.gauss.11.4.562-571
Fitriawan, R., Nadriati, S., & Muharmi, Y. (2021). Sistem Informasi Penilaian Kinerja Guru Berbasis Web Di Smp Negeri 3 Kempas. 1(1), 7–13.
Gusti, G. P. (2023). Malaysian E Commerce Journal ( MECJ ) DIGITAL COMMERCE TRANSFORMATION : A STUDY ON THE IMPACT OF E-MONEY , E-WALLET , AND E-COMMERCE USE ON CONSUMER AND FINANCIAL BEHAVIOR. 7(1), 50–53. https://doi.org/10.26480/mecj.01.2023.50.53
Prasetyo, H., & Waluyo, B. (n.d.). REVITALIZING THE ROLE OF CONSUMERS AND OJK IN OVERCOMING ILLEGAL ONLINE LOAN. 188–197. https://doi.org/10.5281/zenodo.7313458
Rahayu, A. S., Fauzi, A., & Rahmat, R. (2022). Komparasi Algoritma Naïve Bayes Dan Support Vector Machine (SVM) Pada Analisis Sentimen Spotify. Jurnal Sistem Komputer Dan Informatika (JSON), 4(2), 349. https://doi.org/10.30865/json.v4i2.5398
Ramadan, M. F., Dikananda, A. R., & Rifa, A. (2025). Comparison of Sentiment Analysis Models Enhanced by Naïve Bayes and Support Vector Machine Algorithms on Mobile Banking BRImo Reviews. 4(2).
Rika Widianita, D. (2023). No 主観的健康感を中心とした在宅高齢者における 健康関連指標に関する共分散構造分析Title. AT-TAWASSUTH: Jurnal Ekonomi Islam, VIII(I), 1–19.
Setiawan, T., Liem, S., & Pribadi, D. M. R. (2024). Perbandingan Algoritma SVM dan Naïve Bayes dalam Analisis Sentimen Komentar Tiktok pada Produk Skincare. Applied Information Technology and Computer Science, 3(2), 28–32. https://jurnal.politap.ac.id/index.php/aicoms
Syahputra, R., Yanris, G. J., & Irmayani, D. (2022). SVM and Naïve Bayes Algorithm Comparison for User Sentiment Analysis on Twitter. Sinkron, 7(2), 671–678. https://doi.org/10.33395/sinkron.v7i2.11430
Tjikdaphia, N. B. B., & Sulastri, S. (2023). Comparison of Nbc, Svm, Knn Classification Results in Sentiment Analysis of Mobile Jkn. JURTEKSI (Jurnal Teknologi Dan Sistem Informasi), 9(4), 665–672. https://doi.org/10.33330/jurteksi.v9i4.2539
Yoga Pratama, A., Umaidah, Y., Voutama, A., Informatika, T., Ilmu Komputer, F., Singaperbangsa Karawang Ds Paseurjaya, U., Telukjambe Timur, K., Karawang, K., & Barat, J. (2021). Analisis Sentimen Media Sosial Twitter Dengan Algoritma K-Nearest Neighbor Dan Seleksi Fitur Chi-Square (Kasus Omnibus Law Cipta Kerja). Jurnal Sains Komputer & Informatika (J-SAKTI, 5(2), 897–910.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 JOURNAL SAINS STUDENT RESEARCH
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
