Aziz, Muhammad Rafly (2025) ANALISIS SENTIMEN PENGGUNAAPLIKASI OCTO MOBILE BY CIMB NIAGA DI GOOGLE PLAY STORE MENGGUNAKANLATENT DIRICHLET ALLOCATION (LDA) DAN SUPPORT VECTOR MACHINE (SVM). S1 thesis, Universitas Mercu Buana Jakarta - Menteng.
![[img]](https://repository.mercubuana.ac.id/style/images/fileicons/text.png) |
Text (Cover)
41821010045 - Muhammad Rafly Aziz - 01 Cover - Rafly Aziz.pdf Download (322kB) |
|
Text (Bab i)
41821010045 - Muhammad Rafly Aziz - 02 Bab 1 - Rafly Aziz.pdf Restricted to Registered users only Download (285kB) |
|
Text (Bab ii)
41821010045 - Muhammad Rafly Aziz - 03 Bab 2 - Rafly Aziz.pdf Restricted to Registered users only Download (473kB) |
|
Text (Bab iii)
41821010045 - Muhammad Rafly Aziz - 04 Bab 3 - Rafly Aziz.pdf Restricted to Registered users only Download (205kB) |
|
Text (Bab iv)
41821010045 - Muhammad Rafly Aziz - 05 Bab 4 - Rafly Aziz.pdf Restricted to Registered users only Download (1MB) |
|
Text (Bab v)
41821010045 - Muhammad Rafly Aziz - 06 Bab 5 - Rafly Aziz.pdf Restricted to Registered users only Download (129kB) |
|
Text (Daftar Pustaka)
41821010045 - Muhammad Rafly Aziz - 07 Daftar Pustaka - Rafly Aziz.pdf Restricted to Registered users only Download (207kB) |
|
Text (Lampiran)
41821010045 - Muhammad Rafly Aziz - 08 Lampiran - Rafly Aziz.pdf Restricted to Registered users only Download (645kB) |
Abstract
Penurunan rating aplikasi Octo Mobile by CIMB Niaga di Google Play Store menunjukkan adanya ketidakpuasan pengguna, namun permasalahan utamanya adalah belum teridentifikasinya penyebab spesifik di balik penurunan tersebut. Penelitian ini menawarkan solusi dengan menerapkan pendekatan analisis gabungan: menggunakan Support Vector Machine (SVM) untuk mengklasifikasikan sentimen secara akurat, dan Latent Dirichlet Allocation (LDA) untuk mengekstrak topik utama sebagai akar penyebab keluhan. Metode yang dijalankan meliputi perbandingan enam algoritma pada 22.389 ulasan melalui validasi silang 10-fold, yang dilanjutkan dengan pemodelan topik. Hasilnya, SVM terbukti sebagai model terbaik dengan akurasi 95%, dan LDA berhasil mendiagnosis keluhan utama pada topik “Aksesibilitas Akun” dan “Fungsionalitas Aplikasi” yang frekuensinya melonjak pada versi 3.0 dan 3.1. Kombinasi metode ini terbukti efektif memberikan diagnosis masalah yang jelas dan wawasan yang dapat ditindaklanjuti oleh pengembang. The decline in the rating of the Octo Mobile by CIMB Niaga application on the Google Play Store indicates user dissatisfaction, but the main problem is that the specific cause behind the decline has not been identified. This research offers a solution by applying a combined analysis approach: using Support VectorMachine (SVM) to accurately classify sentiment, and Latent Dirichlet Allocation (LDA) to extract key topics as the root cause of complaints. The method involved comparing the six algorithms on 22,389 reviews through 10-fold cross-validation, followed by topic modeling. As a result, SVM proved to be the best model with 95% accuracy, and LDA successfully diagnosed major complaints on the topics “Account Accessibility” and “App Functionality” whose frequency spiked in versions 3.0 and 3.1. This combination of methods proved effective in providing clear problem diagnosis and actionable insights for developers.
| Item Type: | Thesis (S1) |
|---|---|
| NIM/NIDN Creators: | 41821010045 |
| Uncontrolled Keywords: | Analisis Sentimen, Support Vector Machine, Latent Dirichlet Allocation, Ulasan Pengguna, Octo Mobile by CIMB Niaga Sentiment Analysis, Support VectorMachine, Latent Dirichlet Allocation, User Reviews, Octo Mobile by CIMB Niaga |
| Subjects: | 000 Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 000. Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 003 Systems/Sistem-sistem |
| Divisions: | Fakultas Ilmu Komputer > Sistem Informasi |
| Depositing User: | Maulana Arif Hidayat |
| Date Deposited: | 12 Aug 2025 08:44 |
| Last Modified: | 12 Aug 2025 08:44 |
| URI: | http://repository.mercubuana.ac.id/id/eprint/96797 |
Actions (login required)
 |
View Item |
