Claim Missing Document
Check
Articles

Found 4 Documents
Search

Metode Pembobotan Jarak dengan Koefisien Variasi untuk Mengatasi Kelemahan Euclidean Distance pada Algoritma k-Nearest Neighbor Agustiyar Agustiyar; Romi Satria Wahono; Catur Supriyanto
Jurnal Ilmiah SINUS Vol 20, No 1 (2022): Vol. 20 No. 1, Januari 2022
Publisher : STMIK Sinar Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30646/sinus.v20i1.565

Abstract

k-Nearest Neighbor (k-NN) is one of the classification algorithms which becomes top 10 in data mining. k-NN is simple and easy to apply. However, the classification results are greatly influenced by the scale of the data input. All of its attributes are considered equally important by Euclidean distance, but inappropriate with the relevance of each attribute. Thus, it makes classification results decreased. Some of the attributes are more or less relevance or, in fact, irrelevant in determining the classification results. To overcome the disadvantage of k-NN, Zolghadri, Parvinnia, and John proposed Weighted Distance Nearest Neighbor (WDNN) having the performance better than k-NN. However, when the result is k >1, the performance decrease. Gou proposed Dual Distance Weighted Voting k-Nearest Neighbor (DWKNN) having the performance better than k-NN. However, DWKNN focused in determining label of classification result by weighted voting. It applied Euclidean distance without attribute weighting. This might cause all attribute considered equally important by Euclidean distance, but inappropriate with the relevance of each attribute, which make classification results decreased. This research proposed Coefficient of Variation Weighting k-Nearest Neighbor (CVWKNN) integrating with MinMax normalization and weighted Euclidean distance. Seven public datasets from UCI Machine Learning Repository were used in this research. The results of Friedman test and Nemenyi post hoc test for accuracy showed CVWKNN had better performance and significantly different compared to k-NN algorithm. 
Prediksi Penyakit Jantung Menggunakan Attribute Weighting k-Nearest Neighbor Agustiyar Agustiyar
InComTech : Jurnal Telekomunikasi dan Komputer Vol 13, No 2 (2023)
Publisher : Department of Electrical Engineering

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.22441/incomtech.v13i2.17883

Abstract

Penyakit kardiovaskular atau lebih dikenal dengan penyakit jantung menjadi salah satu penyebab kematian tertinggi di Indonesia dan di tingkat global. Selain pola hidup sehat untuk mencegah penyakit tersebut, deteksi dini terhadap resiko penyakit jantung dapat dilakukan dengan data mining atau machine learning salah satunya k-NN. k-NN adalah salah satu metode data mining paling sederhana dan kuat dalam konsistensi hasil klasifikasi, akan tetapi memiliki kekurangan yaitu memberikan bobot yang sama kepada semua atribut. Penelitian ini mengusulkan pembobotan pada atribut untuk mengatasi kelemahan tersebut. Prediksi penyakit jantung digunakan untuk menggambarkan kinerja metode usulan. Pada penelitian ini menggunakan dataset Heart Disease, sebuah dataset publik dari University of California Irvine. Dengan menggunakan nilai k 3, 5, 7, 9 diperoleh rata-rata kinerja metode usulan sebesar 79,87% lebih baik dibandingkan Chi-Square k-NN 79,08% dan k-NN klasik 65,89%. Penelitian ini menyimpulkan bahwa metode pembobotan atribut berhasil mengatasi kekurangan k-NN, jadi metode usulan cocok untuk prediksi penyakit jantung.
A Robust and Interpretable Ensemble Learning Framework for Early Mortality Risk Stratification in Heart Failure Agustiyar Agustiyar; Roman Tsarev; Elham Pishgar
Journal of Intelligent Computing & Health Informatics Vol 7, No 1 (2026): March
Publisher : Universitas Muhammadiyah Semarang Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26714/jichi.v7i1.20873

Abstract

Heart failure remains a formidable global health challenge, frequently complicated by cardiorenal syndrome, which necessitates early and dynamic mortality risk stratification. Existing clinical scoring systems fail to capture complex nonlinear biomarker interactions, whereas state of the art deep learning models suffer from high computational overhead, algorithmic opacity, and a propensity for severe overfitting on small scale tabular data. To address these critical gaps, this study proposes a computationally efficient and transparent ensemble machine learning framework utilizing Extreme Gradient Boosting (XGBoost) coupled with Shapley Additive Explanations (SHAP). The methodology was rigorously benchmarked against a comprehensive suite of ten distinct machine learning architectures using the publicly validated UCI Heart Failure Clinical Records dataset. Comprehensive evaluations demonstrate that the XGBoost framework is both statistically robust and computationally superior. An ablation study confirmed that synthetic resampling critically maximized diagnostic recall across all evaluated models to prevent fatal misclassifications. Calibrated to an optimal clinical decision threshold (τ = 0.35), the model effectively balanced sensitivity with specificity. Furthermore, five fold cross validation confirmed its supreme generalization stability, achieving a peak mean Area Under the Curve of 0.907 with the lowest algorithmic variance (±0.027) among the ten evaluated models, successfully highlighting the vulnerability of unregularized neural networks and classical algorithms on restricted medical datasets. Game theoretic SHAP analysis biologically validated the framework by decisively isolating elevated serum creatinine and reduced ejection fraction as the primary drivers of mortality. By amalgamating sub four millisecond inference latency with mathematically rigorous clinical interpretability, the proposed framework provides a robust, real time screening tool to optimize proactive interventions in acute cardiovascular care.
Face Recognition for Attendance Systems: A Bibliometric Review of Research Trends and Opportunities Agustiyar Agustiyar; R. Rizal Isnanto; Catur Edi Widodo
Jurnal Sisfokom (Sistem Informasi dan Komputer) Vol. 15 No. 1 (2026): JANUARY
Publisher : ISB Atma Luhur

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32736/sisfokom.v15i01.2529

Abstract

Automated attendance systems have become a critical component of smart education environments. This study presents a bibliometric analysis of research on facial recognition-based attendance systems to identify research trends, collaboration patterns, and potential directions for future studies. Data were collected from the Scopus database for the period 2019–2024 using keywords related to “facial recognition,” “attendance system,” and “deep learning.” The bibliometric analysis was conducted using OpenRefine for data cleaning and Biblioshiny (R-Bibliometrix) for visualization and mapping of scientific networks, including co-authorship, keyword co-occurrence, and citation analysis. The results show a significant increase in research publications, dominated by contributions from India, Indonesia, and Malaysia, with deep learning and convolutional neural networks (CNN) as the most frequently studied techniques. International collaboration remains limited, indicating opportunities for broader cooperation in this field. This research contributes by providing a comprehensive overview of the global research landscape on facial recognition for attendance systems and offering strategic insights for developing more accurate, efficient, and scalable recognition technologies in educational environments.