HARYANTI, JESICA WALANDA (2025) ENSEMBLE MODEL NAIVE BAYES DAN RANDOM FOREST UNTUK KLASIFIKASI SENTIMEN DAN PREDIKSI ULASAN PENGGUNA APLIKASI GOOGLE GEMINI. S1 thesis, Universitas Mercu Buana Jakarta.
|
Text (HAL COVER)
01. COVER.pdf Download (316kB) | Preview |
|
![[img]](https://repository.mercubuana.ac.id/style/images/fileicons/text.png) |
Text (BAB I)
02. BAB 1.pdf Restricted to Registered users only Download (242kB) |
|
|
Text (BAB II)
03. BAB 2.pdf Restricted to Registered users only Download (445kB) |
|
|
Text (BAB III)
04. BAB 3.pdf Restricted to Registered users only Download (198kB) |
|
|
Text (BAB IV)
05. BAB 4.pdf Restricted to Registered users only Download (393kB) |
|
|
Text (BAB V)
06. BAB 5.pdf Restricted to Registered users only Download (99kB) |
|
|
Text (DAFTAR PUSTAKA)
07. DAFTAR PUSTAKA.pdf Restricted to Registered users only Download (305kB) |
|
|
Text (LAMPIRAN)
08. LAMLPIRAN.pdf Restricted to Registered users only Download (178kB) |
Abstract
This study aims to analyze user sentiment toward the Google Gemini application on the Google Play Store and to evaluate daily perception trends using a machine learning approach. The variable examined is user sentiment, categorized into positive and negative classes. A total of 3,000 reviews were collected using purposive sampling through web scraping with Python on Google Colaboratory. The data underwent preprocessing steps including text normalization, tokenization, stopword removal, and stemming. The initial classification used the Multinomial Naive Bayes algorithm, enhanced through data balancing with the SMOTE technique and an ensemble model combining Naive Bayes and Random Forest with soft voting. The Naive Bayes model achieved 80% accuracy but had low recall for negative reviews. The ensemble model improved accuracy to 85% and reached a macro F1-score of 0.80. Furthermore, the daily sentiment trend over the past month was successfully modeled with a prediction accuracy of 85.3%. The study concludes that ensemble and data balancing methods are effective in improving sentiment classification performance and are suitable for real-time user opinion monitoring. Keywords: Sentiment Analysis, Google Gemini, Naive Bayes, Ensemble Penelitian ini bertujuan untuk menganalisis sentimen pengguna terhadap aplikasi Google Gemini di Google Play Store dan mengevaluasi tren persepsi harian menggunakan pendekatan machine learning. Variabel yang diteliti adalah sentimen pengguna, dikategorikan menjadi positif dan negatif. Sebanyak 3.000 ulasan dikumpulkan menggunakan metode purposive sampling melalui teknik web scraping dengan Python di Google Colaboratory. Data diproses melalui tahapan preprocessing mencakup normalisasi teks, tokenisasi, penghapusan stopword, dan stemming. Algoritma Multinomial Naive Bayes digunakan untuk klasifikasi awal, kemudian ditingkatkan melalui teknik SMOTE untuk penyeimbangan data dan model ensemble berbasis soft voting yang menggabungkan Naive Bayes dan Random Forest. Hasil analisis menunjukkan bahwa model Naive Bayes menghasilkan akurasi 80% namun kurang sensitif terhadap kelas negatif. Model ensemble meningkatkan akurasi menjadi 85% dan F1-score makro sebesar 0.80. Tren sentimen harian selama satu bulan terakhir juga berhasil dimodelkan dengan akurasi prediksi 85,3%. Kesimpulan penelitian ini menunjukkan bahwa pendekatan ensemble dan balancing data efektif dalam meningkatkan performa klasifikasi dan relevan untuk pemantauan opini pengguna secara berkala. Kata Kunci: Analisis Sentimen, Google Gemini, Naive Bayes, Ensemble
Actions (login required)
 |
View Item |
