Claim Missing Document
Check
Articles

Found 3 Documents
Search

IMPLEMENTASI ARTIFICIAL NEURAL NETWORK PADA KASUS REGRESI LINEAR BERGANDA UNTUK MEMPREDIKSI JUMLAH PAKAN AYAM PETELUR Ali Asgar Zainal Abidin; Kusrini; Ferry Wahyu Wibowo
Jurnal INSTEK (Informatika Sains dan Teknologi) Vol 9 No 2 (2024): OCTOBER
Publisher : Department of Informatics Engineering, Faculty of Science and Technology, Universitas Islam Negeri Alauddin, Makassar, Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.24252/instek.v9i2.50836

Abstract

Produksi telur ayam petelur adalah bagian penting dalam industri peternakan dan berperan besar dalam memenuhi kebutuhan masyarakat akan telur sebagai sumber makanan. Penelitian ini menggunakan algoritma Jaringan Saraf Tiruan (JST), yang sering digunakan untuk memprediksi data, untuk melakukan prediksi jumlah pakan yang dibutuhkan oleh ayam petelur. Penelitian ini bukan tentang hasil prediksi konkret, tetapi lebih tentang evaluasi kemampuan algoritma JST dalam melakukan prediksi berdasarkan dataset yang diperoleh dari sumber Kaggle.Dalam penelitian ini, berbagai model arsitektur jaringan neural telah dieksplorasi, termasuk model dengan struktur 5-30-1, 5-40-1, 5-50-1, dan 5-60-1. Setiap model telah dilatih dan diuji, dan hasilnya menunjukkan bahwa model arsitektur terbaik adalah yang memiliki struktur 5-40-1, dengan tingkat kinerja (MAPE) terendah sekitar 4.04 dan RMSE sebesar 6.71, yang merupakan tingkat kesalahan terendah dibandingkan dengan enam model lainnya. Ini mengindikasikan bahwa model tersebut dapat digunakan dengan baik untuk melakukan prediksi jumlah pakan yang dibutuhkan oleh ayam petelur.
Revisiting Resampling Strategies under Extreme Class Imbalance: Evidence from Large-Scale Online Payment Fraud Detection Ardiansyah, Mursyid; Abidin, Ali Asgar Zainal
Jurnal Pendidikan Informatika (EDUMATIC) Vol 10 No 1 (2026): Edumatic: Jurnal Pendidikan Informatika
Publisher : Universitas Hamzanwadi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29408/edumatic.v10i1.33272

Abstract

Extreme class imbalance in online payment fraud detection creates an accuracy paradox and an operational risk in which improving fraud capture can generate costly false alarms. This study uses a quantitative, experiment-based design to evaluate the operational impact of common resampling strategies under extreme skew using interpretable linear decision rules. The Online Payments Fraud dataset (6.36 million transactions) from Kaggle is analysed using six monetary balance/amount variables (amount, oldbalanceOrg, newbalanceOrig, oldbalanceDest, newbalanceDest) plus the rule-based isFlaggedFraud indicator to predict the isFraud label. Five training variants (no resampling, ROS, RUS, SMOTE, ADASYN) are compared with two linear decision rules: an ordinary least squares linear scoring model (thresholded at 0.5) and a linear SVM, using a leakage-free protocol in which resampling is applied only to the 80% training split and performance is assessed on an untouched, highly imbalanced 20% test set. The findings indicate that LinReg–RUS achieves the most balanced operating point (Precision 65.938%, Recall 47.718%, F1 55.367%, ROC-AUC 98.720%), whereas ADASYN increases recall but collapses precision (~2.1%), yielding F1 ≈4.17%. These results contribute controlled, large-scale evidence that under extreme imbalance, simpler resampling–model combinations can provide more deployable precision–recall trade-offs than aggressive synthetic sampling, supporting interpretable baselines for capacity-constrained payment screening.
Implementation of a Deep Learning Model for Real-Time Detection and Classification of Toraja Traditional Motifs (Pa’ssura’) for Digital Cultural Preservation Abidin, Ali Asgar Zainal; Ardiansyah, Mursyid; Zahra, Aqilah
Sistemasi: Jurnal Sistem Informasi Vol 15, No 4 (2026): Sistemasi: Jurnal Sistem Informasi
Publisher : Program Studi Sistem Informasi Fakultas Teknik dan Ilmu Komputer

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32520/stmsi.v15i4.6132

Abstract

Toraja traditional motifs, known as Pa’ssura’, represent an intangible cultural heritage rich in philosophical values and deep cultural identity. However, the authenticity and understanding of the meanings behind these motifs are at risk of erosion among younger generations due to the lack of interactive and technologically relevant learning media. This study aims to bridge this gap through an innovative digital cultural preservation strategy by implementing deep learning technology. Specifically, the research focuses on developing a real-time object detection and classification system using a Convolutional Neural Network (CNN) architecture, particularly the YOLO11s model. The main research stages include constructing an annotated image dataset for seven primary Pa’ssura’ motifs: Pa’ Barre Allo, Pa’ Kapu Baka, Pa’ Tangke Lumu, Pa’ Tedong, Pa’ Ulu Karua, Pa’ Kadang Pao, and Pa’ Papan Kandaure. These data were collected from both planar media (such as textiles) and non-planar media, including wood carvings and stone engravings. The results show that the developed model achieved a precision of 0.7109, a recall of 0.6708, and an mAP@50 of 0.6910 after 100 training epochs. The implementation of data augmentation techniques proved effective in increasing the dataset size—from 1,050 images before augmentation to 2,520 images after augmentation—thereby significantly enhancing the model’s robustness in detecting and classifying motifs across both planar and non-planar media. This study produces an accurate and practical model that can be applied as an educational tool in mobile applications. Furthermore, the model plays an important role in preserving Toraja cultural heritage through digitalization.