Article Per Year (5 Year)
p-Index From 2021 - 2026
0.562
P-Index
This Author published in this journals
All Journal Jurnal Sains dan Informatika IKRAM: Jurnal Ilmu Komputer Al Muslim Jurnal Inkofar
Dyah Ayu Nurmumpuni
Universitas Pamulang
Computer Science & IT Decision Sciences, Operations Research & Management Engineering Mathematics

Published : 3 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 3 Documents
Search

 1 
Analisis Perbandingan Metode Naïve Bayes dan Decision Tree C4.5 untuk Meningkatkan Kualitas Sampling dan Efisiensi Biaya (Studi Kasus: PT. BFI Finance Tbk) Dyah Ayu Nurmumpuni; Tukiyat Tukiyat; Achmad Hindasyah
Jurnal Sains dan Informatika Vol. 12 No. 1 (2026): Jurnal Sains dan Informatika
Publisher : Teknik Informatika, Politeknik Negeri Tanah Laut

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.34128/jsi.v12i1.1744

Abstract

Penelitian ini bertujuan menganalisis komparasi metode Naïve Bayes dan Decision Tree C4.5 dalam mengoptimalkan proses sampling audit pada PT BFI Finance Tbk. Studi dilakukan melalui pendekatan kuantitatif dengan menggunakan dataset 491 kontrak, terdiri dari 461 data latih dan 30 data uji. Variabel independen mencakup angsuran, lama hari menunggak, sisa pokok hutang, dan jenis produk, sedangkan variabel dependen adalah hasil temuan audit. Metode penelitian melibatkan implementasi algoritma machine learning untuk mengklasifikasikan kualitas sampling. Hasil penelitian menunjukkan bahwa Decision Tree C4.5 memiliki akurasi lebih tinggi dibandingkan Naïve Bayes, dengan kemampuan superior dalam menangani kompleksitas data. Evaluasi performa algoritma mencakup metrik akurasi, presisi, recall, dan F1-score, yang mengindikasikan keunggulan Decision Tree C4.5 dalam meningkatkan efisiensi proses audit.
Prediksi Tingkat Retensi Pengguna Aplikasi Digital Menggunakan Artificial Neural Network Berbasis Data Aktivitas Pengguna Dyah Ayu Nurmumpuni; Novi Wulandari; Angelina Hadriani
IKRAM: Jurnal Ilmu Komputer Al Muslim Vol. 5 No. 1 (2026): IKRAM: Jurnal Ilmu Komputer Al Muslim
Publisher : STMIK Al Muslim

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

User retention is a critical indicator of success for digital applications, particularly in highly competitive technology-driven markets. This study aims to predict user retention using an Artificial Neural Network (ANN) approach based on user activity data. The dataset utilized is obtained from Kaggle and includes behavioral features such as usage frequency, session duration, and user interactions within the application. The research methodology involves data preprocessing, ANN model development with multiple architectural variations, and evaluation using metrics including Accuracy, Precision, Recall, F1-Score, Confusion Matrix, and ROC-AUC. The results demonstrate that the ANN model achieves an accuracy of approximately 88%, precision of 90%, recall of 97%, and an F1-score of 93%, with an AUC value above 0.90, indicating strong classification performance. However, further analysis reveals signs of overfitting and a relatively high number of false positives due to data imbalance. Additionally, increasing model complexity does not necessarily lead to improved performance. Overall, ANN proves to be an effective approach for predicting user retention, although further optimization is required to achieve a more balanced and robust model for real-world implementation.
Design and Implementation of an Android-Based Student Attendance Application Utilizing QR Codes to Enhance Presence Efficiency Dyah Ayu Nurmumpuni; Angelina Hadriani
INKOFAR Vol. 10 No. 1 (2026)
Publisher : Politeknik META Industri Cikarang

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Background Manual student attendance recording in educational institutions is inefficient, error prone, and susceptible to proxy attendance (titip absen). Physical presence recapitulation also delays real time attendance reporting for administrators and parents. Purpose This study designs, implements, and evaluates an Android based attendance information system incorporating dual factor verification through dynamic QR codes and location based geofencing. Methodology The system combines dynamically rotating QR tokens (60 second expiration) with the Haversine formula for real time GPS distance validation within a 50M radius boundary. System performance and user acceptance were evaluated using Black box testing, UAT, MOS surveys (N = 5 teachers, N = 30 students), response latency benchmarking, and server load stress testing up to 100 concurrent requests. Findings Empirical evaluation demonstrates that the proposed system reduces classroom attendance duration from an average of 12.5 minutes (10 to 15 minutes) to 1.4 minutes per class session, achieving an 88.8% reduction in administrative time. The average MOS satisfaction score reached 4.57/5.00 for teachers and 4.62/5.00 for students (Excellent category). Response latency benchmarks yielded an average response time of 245 ms under standard operational conditions, maintaining a 0.0% error under load tests up to 50 concurrent requests. Implications The solution offers primary educational institutions a cost effective attendance verification without requiring dedicated door terminal hardware, while providing guidelines for managing indoor GPS signal drift. Originality The primary scientific contribution lies in integrating short lived cryptographic QR tokens with GPS geofencing on commodity mobile devices, providing a zero hardware, zero proxy presence verification mechanism tailored for primary schools.
Co-Authors Achmad Hindasyah Angelina Hadriani Angelina Hadriani Novi Wulandari Tukiyat Tukiyat