Article Per Year (5 Year)
p-Index From 2021 - 2026
0.444
P-Index
This Author published in this journals
All Journal Journal of Computers and Digital Business Journal of Intelligent Software Systems
Ivónia Fátima Ruas da Silva
Universitas Teknologi Digital Indonesia
Computer Science & IT Decision Sciences, Operations Research & Management

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

 1 
Studi Eksploratif Pipeline Multilayer Perceptron pada Dataset Sintetik Berlabel Deterministik: Implikasi Metodologis untuk Klasifikasi Hipertensi Ivónia Fátima Ruas da Silva; Bambang Purnomosidi Dwi Putranto; Widyastuti Andriyani
Journal of Computers and Digital Business Vol. 5 No. 2 (2026)
Publisher : PT. Delitekno Media Mandiri

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.56427/jcbd.v5i2.991

Abstract

Hipertensi merupakan penyakit kardiovaskular dengan prevalensi tinggi dan menjadi penyebab utama mortalitas global, sehingga deteksi dini menjadi kebutuhan klinis yang krusial. Namun, pengembangan model deep learning pada konteks sumber daya terbatas sering terkendala ketersediaan dataset berskala besar. Gap penelitian yang diidentifikasi adalah belum tersedianya studi eksploratif yang secara eksplisit menguji kelayakan pipeline Multilayer Perceptron (MLP) sederhana pada dataset berukuran sangat kecil dengan dokumentasi reproducible. Penelitian ini bertujuan mendemonstrasikan pipeline MLP end-to-end pada dataset sintetik 150 sampel dengan sembilan fitur biometrik dan gaya hidup. Setelah one-hot encoding dan normalisasi Min-Max, dimensi input menjadi 15 neuron. Arsitektur MLP terdiri atas tiga hidden layer (64-32-16, ReLU) dan output sigmoid, dilatih 100 epoch menggunakan optimizer Adam (learning rate 0,001; batch size 16) dengan early stopping. Evaluasi pada test set (n = 30) memperoleh akurasi 90,00%, presisi 85,00%, recall 100%, F1-score 91,90%, dan AUC-ROC 0,91, dengan tiga false positive teridentifikasi sebagai kasus borderline pre-hypertension. Kontribusi penelitian terletak pada penyajian artefak reproducible—dataset sintetik, kode preprocessing, dan visualisasi diagnostik—sebagai baseline pedagogis untuk institusi berketerbatasan data. Keterbatasan utama, yaitu sifat deterministik label yang berpotensi menimbulkan circular reasoning pada fitur tekanan darah, didokumentasikan eksplisit sebagai catatan validitas internal.
DIGITAL ACTIVITY LOCATION CLUSTERING BASED ON TWITTER GEOSPATIAL DATA FOR SPATIOTEMPORAL BUSINESS INTELLIGENCE Triyan Agung Laksono; Widyastuti Andriyani; Fadhlih Girindra Putra; Ivonia Fatima Ruas da silva; Wiwi widayani
Journal of Intelligent Software Systems Vol 4, No 1 (2025): Juli 2025
Publisher : LPPM UTDI (d.h STMIK AKAKOM) Yogyakarta

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26798/jiss.v4i1.2005

Abstract

This research develops an approach for clustering digital activity locations based on Twitter geospatial data with the aim of supporting business intelligence spatiotemporal . By utilizing the Twitter Geospatial Data dataset containing more than 14 million tweets geo-tagged from the United States, this study implements and compares the DBSCAN and K- Means algorithms to identify spatial and temporal patterns of Twitter user activity. The research process begins with the data pre -processing stage using the Knowledge Discovery Database (KDD), followed by the implementation of the clustering algorithm , and ending with the integration of the results into the dashboard.business intelligence using Power BI . The results show that DBSCAN is able to detect irregular clusters that follow geographic patterns and population density, while K- Means produces a division of the region into three main clusters (West Coast, Central Region, and East Coast) with different temporal activity patterns. Integration of clustering results into a BI dashboard produces actionable business insights , such as identification of digital activity hotspots , optimal time for content delivery, geographic segmentation for marketing strategies, and temporal activity patterns for campaign scheduling. This research contributes to the development of an integrated spatiotemporal analysis pipeline to support data-driven decision making.
Co-Authors Andriyani, Widyastuti Bambang Purnomosidi Dwi Putranto Fadhlih Girindra Putra Triyan Agung Laksono Wiwi Widayani