THERESA, OKTANIA GERALDINE (2025) ANALISIS SENTIMEN APLIKASI ISAKU PADA ULASAN PENGGUNA DI GOOGLE PLAY STORE MENGGUNAKAN ALGORITMA NAIVE BAYES. S1 thesis, Universitas Mercu Buana Jakarta.
|
Text (HAL COVER)
01 COVER.pdf Download (741kB) | Preview |
|
![]() |
Text (BAB I)
02 BAB 1.pdf Restricted to Registered users only Download (245kB) |
|
![]() |
Text (BAB II)
03 BAB 2.pdf Restricted to Registered users only Download (406kB) |
|
![]() |
Text (BAB III)
04 BAB 3.pdf Restricted to Registered users only Download (288kB) |
|
![]() |
Text (BAB IV)
05 BAB 4.pdf Restricted to Registered users only Download (645kB) |
|
![]() |
Text (BAB V)
06 BAB 5.pdf Restricted to Registered users only Download (145kB) |
|
![]() |
Text (DAFTAR PUSTAKA)
07 DAFTAR PUSTAKA.pdf Restricted to Registered users only Download (267kB) |
|
![]() |
Text (LAMPIRAN)
08 LAMPIRAN.pdf Restricted to Registered users only Download (413kB) |
Abstract
This study aims to analyze user sentiment toward the iSaku application using the Multinomial Naïve Bayes algorithm. A total of 1,200 Indonesianlanguage user reviews were collected via web scraping from the Google Play Store. The preprocessing steps included case folding, tokenizing, stopword removal, stemming, and text cleaning. The textual data was then transformed into numerical representation using the TF-IDF (Term Frequency–Inverse Document Frequency) method. To address class imbalance among sentiment categories (positive, neutral, negative), the SMOTE technique was applied to the training data. Model training was conducted using a pipeline combined with parameter tuning through GridSearchCV. Evaluation results showed that accuracy increased from 55% to 78% after applying SMOTE. Notably, the F1-score for the neutral class improved significantly. This research demonstrates that the combination of TF-IDF, Multinomial Naïve Bayes, and SMOTE can effectively classify short user-generated text. The findings can support developers in monitoring user feedback and improving iSaku's service quality. Keywords: Sentiment Analysis, Naïve Bayes, TF-IDF, SMOTE, iSaku Penelitian ini bertujuan untuk menganalisis sentimen ulasan pengguna aplikasi iSaku menggunakan algoritma Multinomial Naïve Bayes. Sebanyak 1.200 data ulasan berbahasa Indonesia dikumpulkan melalui teknik web scraping dari Google Play Store. Proses preprocessing meliputi case folding, tokenizing, stopword removal, stemming, dan cleaning. Data teks dikonversi menjadi representasi numerik menggunakan metode TF-IDF. Untuk mengatasi ketidakseimbangan kelas sentimen (positif, netral, negatif), diterapkan metode SMOTE pada data latih. Proses pelatihan model dilakukan menggunakan pipeline dan tuning parameter dengan GridSearchCV. Hasil evaluasi menunjukkan akurasi meningkat dari 55% menjadi 78% setelah penerapan SMOTE. Peningkatan juga terjadi pada F1-score kelas netral yang sebelumnya sangat rendah. Penelitian ini menunjukkan bahwa kombinasi TF-IDF, Multinomial Naïve Bayes, dan SMOTE mampu menghasilkan performa klasifikasi yang baik terhadap ulasan teks pendek. Hasil ini dapat digunakan sebagai dasar untuk memahami persepsi pengguna dan meningkatkan layanan aplikasi iSaku. Kata kunci : Analisis Sentimen, Naïve Bayes, TF-IDF, SMOTE, iSaku
Actions (login required)
![]() |
View Item |