Indonesia holds a strategic position in the global manufacturing sector, with a manufacturing output of USD 228.32 billion in 2021, ranking 10th worldwide. In 2023, it ranked 12th globally by manufacturing value added, according to the World Bank’s report. However, this growth is accompanied by 297,725 workplace accidents reported in Indonesia in 2022, marking a 27.03% increase from the previous year. This study aims to develop a Personal Protective Equipment (PPE) monitoring system using Big Data, employing Convolutional Neural Network (CNN) and You Only Look Once (YOLO) algorithms. The dataset consists of at least 1,000 images for each of four classes: Helmet, Vest, NoHelmet, and NoVest. Evaluation results show a mAP@50 of 83.1%, with the highest detection performance in Vest (0.90), followed by NoHelmet (0.88), Helmet (0.85), and NoVest (0.81). These findings demonstrate strong potential in supporting safety protocol compliance and reducing workplace accidents in high-risk industrial environments.
 
Keywords - Big Data, Convolutonal Neural Network, You Only Look Once

Segmentasi pelanggan memainkan peran penting dalam meningkatkan profitabilitas bisnis dengan mengidentifikasi pola perilaku pelanggan. Penelitian ini mengusulkan pendekatan segmentasi berbasis model LRFM (Length, Recency, Frequency, Monetary) untuk meningkatkan ketepatan klasifikasi pelanggan. Dengan menggunakan data transaksi, skor LRFM dihitung dan diklasifikasikan ke dalam empat kategori: Best Customers (121 pelanggan), Loyal Customers, Potential, dan At Risk. Untuk meningkatkan interpretabilitas hasil, tiga teknik reduksi dimensi : t-SNE, UMAP, dan PaCMAP digunakan dalam analisis visual. Di antara ketiganya, UMAP menunjukkan visualisasi paling seimbang berdasarkan indikator separation clarity, cluster compactness, dan outlier identification. Hasil penelitian menunjukkan bahwa segmentasi visual membantu pengambil keputusan dalam mengidentifikasi kelompok pelanggan strategis untuk program retensi dan optimalisasi. Studi ini memberikan referensi berharga bagi industri yang ingin meningkatkan manajemen hubungan pelanggan melalui pendekatan berbasis data.
 
Kata kunci - Model LRFM, profitabilitas bisnis, reduksi dimensi, segmentasi pelanggan, UMAP