Penerapan Algoritma K-Nearest Neighbor Pada Analisis Sentimen Terhadap Ulasan Aplikasi Mybca Di Google Play Store

Rifka Noor Ikhsan, Muhammad (2024) Penerapan Algoritma K-Nearest Neighbor Pada Analisis Sentimen Terhadap Ulasan Aplikasi Mybca Di Google Play Store. S1 Teknik Informatika thesis, STMIK Widya Cipta Dharma.

	Text 2243908-S1- Jurnal.pdf Download (1MB)
	Text 2243908-S1- Teknik Informatika.pdf Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Banyak pengguna mobile banking mengeluhkan kinerja aplikasi myBCA terkait fitur teknis dan fungsional, seperti kegagalan saat top-up, proses transaksi, login, serta aplikasi yang berjalan lambat meskipun koneksi jaringan stabil. Untuk meningkatkan kualitas dan reputasi aplikasi mobile banking myBCA, serta mempertahankan kepercayaan nasabah terhadap keandalan, efektivitas, dan efisiensi aplikasi, diperlukan analisis sentimen sebagai bahan evaluasi bagi PT. Bank Central Asia, Tbk., terutama pada layanan perbankan mobile banking. Penelitian ini menggunakan metode dari Keputusan Menteri Ketenagakerjaan Republik Indonesia Nomor 299 Tahun 2020 tentang Penetapan Standar Kompetensi Kerja Nasional Indonesia (SKKNI). Data diambil melalui scraping menggunakan library google-play-scraper, sedangkan tahap NLP dan preprocessing teks menggunakan library NLP-ID. Proses pelabelan dilakukan menggunakan kamus sentimen dari InSet dan SentiWord. Teknik pembobotan yang digunakan adalah TF-IDF, dan data yang sudah dibobot diimbangi menggunakan SMOTE untuk meningkatkan performa model K-Nearest Neighbors (KNN). Pengklasifikasian data ulasan dilakukan menggunakan model K-Nearest Neighbors, dengan penentuan jumlah tetangga optimal melalui cross-validation, mencari 10 skor terbaik, dan elbow method untuk menentukan rentang terbaik dari 1 hingga 11. Model K-Nearest Neighbors yang dihasilkan memiliki akurasi 81%, dan dievaluasi menggunakan Confusion Matrix. Hasil analisis dan pembangunan model K-Nearest Neighbors diimplementasikan dalam bentuk website menggunakan web service dari Streamlit yang dipadukan dengan web hosting dari GitHub Codespaces. Website ini digunakan untuk menganalisis sentimen terhadap layanan pada aplikasi myBCA di Google Play Store berdasarkan ulasan-ulasan pengguna, serta diuji menggunakan White Box Testing dan Black Box Testing. =========================================================== Many mobile banking users complain about the performance of the myBCA application related to technical and functional features, such as failures during top-ups, transaction processes, logins, and applications that run slowly even though the network connection is stable. To improve the quality and reputation of the myBCA mobile banking application, as well as maintain customer trust in the reliability, effectiveness, and efficiency of the application, sentiment analysis is needed as an evaluation material for PT. Bank Central Asia, Tbk., especially in mobile banking services. This study uses the method from the Decree of the Minister of Manpower of the Republic of Indonesia Number 299 of 2020 concerning the Determination of the Indonesian National Work Competency Standards (SKKNI). Data was taken through scraping using the google-play-scraper library, while the NLP and text preprocessing stages used the NLP-ID library. The labeling process was carried out using sentiment dictionaries from InSet and SentiWord. The weighting technique used was TF-IDF, and the weighted data was balanced using SMOTE to improve the performance of the K-Nearest Neighbors (KNN) model. Review data classification was carried out using the K-Nearest Neighbors model, by determining the optimal number of neighbors through cross-validation, finding the 10 best scores, and the elbow method to determine the best range from 1 to 11. The resulting K-Nearest Neighbors model had an accuracy of 81%, and was evaluated using the Confusion Matrix. The results of the analysis and development of the K-Nearest Neighbors model are implemented in the form of a website using a web service from Streamlit combined with web hosting from GitHub Codespaces. This website is used to analyze sentiment towards services on the myBCA application on the Google Play Store based on user reviews, and is tested using White Box Testing and Black Box Testing.

Item Type:	Thesis (S1 Teknik Informatika)
Additional Information:	Pembimbing 1 : Pitrasacha Adytia, S.T., M.T. Pembimbing 2 : Wahyuni, S.Kom., M.Kom.
Uncontrolled Keywords:	K-Nearest Neighbor, Analisis Sentimen, SKKNI, Layanan
Subjects:	Q Science > QA Mathematics > QA76 Computer software
Divisions:	Teknik Informatika
Depositing User:	Muhammad Rifka Noor Ikhsan
Date Deposited:	07 Aug 2024 06:52
Last Modified:	07 Aug 2024 06:52
URI:	http://repository.wicida.ac.id/id/eprint/5591

Actions (login required)

View Item