Analisis Sentimen Pada Ulasan Aplikasi Pemerintah Kota Surabaya pada Google Play Store Menggunakan Metode Ekstraksi Vektor Kata Dan Metode Ensemble

Fauzan, Mohammad Ahnaf (2025) Analisis Sentimen Pada Ulasan Aplikasi Pemerintah Kota Surabaya pada Google Play Store Menggunakan Metode Ekstraksi Vektor Kata Dan Metode Ensemble. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5025211170-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (11MB)

Abstract

Pemerintah Kota Surabaya terus berupaya meningkatkan layanan publik melalui inovasi digital dengan aplikasi seperti Wargaku Surabaya, Peken Surabaya, Kantorku, dan Sapu Jagat. Aplikasi-aplikasi ini dirancang untuk mempermudah interaksi antara pemerintah dan warga, memperlancar belanja kebutuhan pokok di UMKM lokal, mendukung operasional merchant, serta memantau aktivitas pegawai secara real-time. Analisis ulasan pengguna di platform seperti Google Play Store sangat penting untuk memberikan wawasan tentang kepuasan pengguna dan umpan balik untuk pengembangan aplikasi lebih lanjut.Penelitian ini melakukan analisis sentimen terhadap ulasan pengguna dengan menggunakan model ekstraksi fitur IndoBERT uncased, TF-IDF, dan FastText. Untuk klasifikasi sentimen, digunakan metode ensemble Voting Classifier, Stacking, dan DES KNORA-E, yang menggabungkan Naive Bayes, SVM, dan Logistic Regression. Hasil evaluasi menunjukkan bahwa model base terbaik adalah Logistic Regression menggunakan input hasil ekstraksi vector kata FastText, Logistic Regression menunjukkan kinerja terbaik dengan precision 0.9470, recall 0.9473, dan F1-Score 0.9471. Dalam hal metode ensemble, Logistic Regression dengan FastText memberikan hasil terbaik, namun DES Knora-e dan Stacking Classifier juga memberikan hasil yang sangat baik pada FastText, dengan precision 0.9470, recall 0.9473, dan F1-Score 0.9471. Meskipun metode ensemble seperti DES Knora-e dan Stacking Classifier memberikan hasil identik dengan Logistic Regression pada FastText, Logistic Regression tetap menunjukkan performa terbaik secara konsisten di seluruh skenario dan Soft Voting sebagai ensemble terbaik
=====================================================================================================================================
The Surabaya City Government continues to improve public services through digital innovations with applications such as Wargaku Surabaya, Peken Surabaya, Kantorku, and Sapu Jagat. These applications are designed to facilitate interaction between the government and citizens, streamline the purchase of essential goods from local SMEs, support merchant operations, and monitor employee activities in real-time. Analyzing user reviews on platforms like Google Play Store is crucial for gaining insights into user satisfaction and providing feedback for further application development.This study performs sentiment analysis on user reviews using feature extraction models such as IndoBERT uncased, TF-IDF, and FastText. For sentiment classification, the Voting Classifier, Stacking, and DES-KNORA-E ensemble method, which combines Naive Bayes, SVM, and Logistic Regression.The evaluation results show that the best base model is Logistic Regression using the FastText word vector extraction. Logistic Regression performs the best with a precision of 0.9470, recall of 0.9473, and F1-Score of 0.9471. In terms of ensemble methods, Logistic Regression with FastText provides the best results, while DES Knora-e and Stacking Classifier also yield excellent results on FastText, with a precision of 0.9470, recall of 0.9473, and F1-Score of 0.9471. Although ensemble methods like DES Knora-e and Stacking Classifier provide identical results to Logistic Regression on FastText, Logistic Regression consistently shows the best performance across all scenarios, with Soft Voting being the best ensemble method.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Analisis Sentimen, Aplikasi Pemerintah Surabaya, Ekstraksi Vektor Kata, Metode Ensemble.Sentiment Analysis, Surabaya Government Applications, Word Vector Extraction, Ensemble Methods.
Subjects:	T Technology > T Technology (General)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Mohammad Ahnaf Fauzan
Date Deposited:	30 Jul 2025 06:51
Last Modified:	30 Jul 2025 06:51
URI:	http://repository.its.ac.id/id/eprint/123116

Actions (login required)

View Item