Claim Missing Document
Check
Articles

Found 3 Documents
Search

Model Klastering Hybrid Menggunakan Inisialisasi K-means++ dan Algoritma Optimasi Grey Wolf Mukti, Bayu Priya
JUSTIN (Jurnal Sistem dan Teknologi Informasi) Vol 13, No 2 (2025)
Publisher : Jurusan Informatika Universitas Tanjungpura

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26418/justin.v13i2.88211

Abstract

Penelitian ini mengembangkan GWO-KMeans++, model klastering hybrid yang mengintegrasikan Grey Wolf Optimizer (GWO) dengan K-Means++ untuk mengatasi masalah local optima dalam inisialisasi centroid. Model diuji pada lima dataset UCI (Seeds, Wine, Sonar, Bank, Forest) dengan karakteristik beragam, mulai dari data pertanian berdimensi rendah (6 fitur) hingga sinyal sonar berisik (60 fitur). Kinerja diukur menggunakan Silhouette Score (SC) dan Davies-Bouldin Index (DB) untuk jumlah klaster k=2"“10, lalu dibandingkan dengan K-Means++ melalui Uji Wilcoxon Signed-Rank. Hasil menunjukkan GWO-KMeans++ meningkatkan SC sebesar 19,71"“24,59% (Seeds, k=5"“7), 56,81% (Wine, k=5), dan 210,85% (Sonar, k=2), serta mengurangi DB hingga 22,19% (Seeds, k=7) dan 28,02% (Wine, k=5). Uji statistik mengonfirmasi peningkatan SC signifikan di semua dataset (p 0,05), dengan nilai p=0,0039 (Seeds, Wine, Sonar), p=0,0117 (Bank), dan p=0,0273 (Forest). Namun, perbaikan DB hanya signifikan pada Seeds (p=0,0117) dan Wine (p=0,0078). Visualisasi klaster memperlihatkan distribusi data lebih terpisah dan centroid lebih akurat, khususnya pada data multidimensi (Wine) dan berisik (Sonar). Model ini stabil pada k=3"“6, cocok untuk data nonlinier, dengan aplikasi di bioinformatika hingga deteksi kecurangan keuangan. Rekomendasi lanjutan meliputi optimasi parameter GWO, integrasi reduksi dimensi (PCA), dan pengujian pada dataset big data.
Leveraging TF-IDF and Random Forest to Uncover Genre Patterns in Google Books Metadata Putri, Nadya Awalia; Mukti, Bayu Priya
International Journal for Applied Information Management Vol. 5 No. 4 (2025): Regular Issue: December 2025
Publisher : Bright Institute

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/ijaim.v5i4.112

Abstract

This paper presents a machine learning-based approach for classifying books into genres using their descriptions. We employed a Random Forest classifier combined with Term Frequency-Inverse Document Frequency (TF-IDF) to convert text descriptions into numerical features, enabling the classification of books into six genres: Fiction, Literary Criticism, Education, Social Science, Biography & Autobiography, and Unknown Genre. The model was trained and evaluated on a dataset sourced from Google Books, which was preprocessed to remove missing data and clean the text descriptions by eliminating punctuation, numbers, and stopwords. We performed 5-fold cross-validation to assess the model's performance, which resulted in an average cross-validation accuracy of 64.22%. The final model achieved an accuracy of 62.71% on the test set, with the highest recall observed in the "Fiction" genre. The results indicated that the Random Forest classifier was particularly effective in classifying well-represented genres like "Fiction" and "Unknown Genre." However, genres with fewer samples, such as "Social Science" and "Biography & Autobiography," showed poor performance, highlighting the challenges posed by class imbalance and data sparsity. A confusion matrix and classification report revealed these discrepancies, with certain genres being misclassified more often than others. This research demonstrates the feasibility of using machine learning for automated book genre classification, offering significant potential for enhancing book recommendation systems and improving user experience. Despite its promising results, the study's limitations, including data sparsity and genre imbalance, suggest that further work is needed to refine the model. Future research could explore the use of deep learning techniques and the expansion of the dataset to address these issues and improve genre classification accuracy. The potential for automated genre classification in real-world applications, such as book categorization and personalized recommendations, presents an exciting direction for the book industry.
Fraud-Risk Classification in Indonesian MSME-Related Online Narratives: A Comparison of IndoBERT and TF-IDF–SVM Mukti, Bayu Priya
Journal of Applied Informatics and Computing Vol. 10 No. 5 (2026): October 2026
Publisher : Politeknik Negeri Batam

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30871/jaic.v10i5.13801

Abstract

Digital fraud threatens micro, small, and medium enterprises (MSMEs), yet most automated detection studies remain centered on structured transactions, URLs, emails, or English-language benchmarks. This study evaluates fraud-risk narrative classification in 13,541 deduplicated Indonesian-language texts collected during a three-year window spanning 2023-2026 from Kompas, Detik, YouTube, and X. All retained documents were human-reviewed and assigned a binary discourse-level label and, for risk-positive records, one of seven fraud categories. A tuned TF-IDF linear support vector machine (SVM) was compared with IndoBERT using group-disjoint train-validation-test partitions, validation-only threshold selection, cost-sensitive learning, and 2,000 paired bootstrap resamples. On the untouched test set (n=2,032), IndoBERT achieved 0.9586 Macro-F1, 0.9636 accuracy, 0.9177 Matthews correlation coefficient, and 0.9771 PR-AUC. Its mean Macro-F1 improvement over SVM was 0.0149 with a 95% bootstrap confidence interval of 0.0045-0.0256, whereas the PR-AUC difference was small and uncertain (0.0022; 95% CI: -0.0044-0.0090). IndoBERT reduced false negatives from 69 to 50 and false positives from 31 to 24. Secondary analyses exposed uneven category difficulty, strong topic overlap around social engineering, and a length-sensitive trade-off: IndoBERT was better on short texts but worse than SVM on documents exceeding 512 words. The contribution is an auditable Indonesian MSME-related fraud benchmark that combines duplicate control, a strong classical baseline, calibrated decision thresholds, paired uncertainty, category-level diagnostics, and error analysis rather than proposing a new neural architecture.