Article Per Year (5 Year)
p-Index From 2021 - 2026
0.408
P-Index
This Author published in this journals
All Journal Progresif: Jurnal Ilmiah Komputer
Udjulawa, Daniel
Universitas Multi Data Palembang
Computer Science & IT Control & Systems Engineering

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

Found 2 Documents
Search

 1 
Penerapan Algoritma Random Forest Berbasis Shap Feature Importance dan GridsearchCV Untuk Deteksi Phishing Pratama, Samuel Effendi; Udjulawa, Daniel
Progresif: Jurnal Ilmiah Komputer Vol 22, No 1 (2026): Januari
Publisher : STMIK Banjarbaru

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35889/progresif.v22i1.3345

Abstract

The rapid growth of internet users in Indonesia has increased the risk of cyberattacks, particularly phishing. Phishing is a digital fraud attempt that disguises links to resemble official websites in order to steal users’ sensitive information. This study aims to develop a phishing link detection model using a machine learning approach. The dataset consists of 11,430 URL entries from Mendeley Data, including features such as URL length, suspicious symbols, and subdomain levels. The Random Forest algorithm was chosen for its ability to handle high-dimensional data and resist overfitting. Feature selection was performed using SHAP (Shapley Additive Explanations) to assess feature contributions, while model optimization was conducted with GridSearchCV. The best configuration, RF + GS + SHAP Threshold-P10, achieved an accuracy of 0.9650 and an F1-score of 0.9651, producing an accurate, efficient, and interpretable phishing detection model.Keywords: Phishing; Random Forest; GridSearchCV; SHAP; Machine Learning AbstrakPesatnya pertumbuhan pengguna internet di Indonesia meningkatkan risiko serangan siber, salah satunya phishing. Phishing merupakan upaya penipuan digital dengan menyamarkan tautan agar menyerupai situs resmi untuk mencuri informasi sensitif pengguna. Penelitian ini bertujuan membangun model deteksi tautan phishing menggunakan pendekatan machine learning. Dataset yang digunakan berisi 11.430 entri URL dari Mendeley Data, mencakup fitur seperti panjang URL, simbol mencurigakan, dan tingkat subdomain. Algoritma random forest dipilih karena mampu menangani data berdimensi tinggi serta tahan terhadap overfitting. Seleksi fitur dilakukan dengan SHAP (Shapley Additive Explanations) untuk menilai kontribusi fitur, sedangkan optimasi parameter model menggunakan GridSearchCV. Hasil penelitian menunjukkan konfigurasi RF + GS + SHAP Threshold-P10 memberikan akurasi 0,9650 dan F1-score 0,9651, menghasilkan model yang akurat, efisien, dan transparan dalam mendeteksi tautan phishing.Kata kunci: Phishing; Random Forest; GridSearchCV; SHAP; Machine Learning
Klasifikasi Mutu Biji Kopi Menggunakan Metode CNN-SVM Berdasarkan Cacat Fisik dan Warna Gunawan, Michael; Udjulawa, Daniel
Progresif: Jurnal Ilmiah Komputer Vol 22, No 1 (2026): Januari
Publisher : STMIK Banjarbaru

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35889/progresif.v22i1.3344

Abstract

The determination of coffee bean quality in Indonesia is generally still done manually based on physical defects and color, which is subjective and time-consuming. This study aims to develop a digital image-based green coffee bean quality classification model using the Convolutional Neural Network and Support Vector Machine (CNN-SVM) method. CNN is used as a feature extractor with a ResNet-50 architecture, while SVM functions as a classifier using a Radial Basis Function (RBF) kernel. The dataset consists of 10 classes of coffee bean defects and is divided into 80% training data and 20% test data. The test results show an accuracy value of 77.68%, precision of 80.04%, recall of 77.15%, and f1-score of 77.47%. This approach proves that the combination of CNN and SVM can improve the accuracy and stability of the model. This finding is a novelty in the development of an efficient and objective artificial intelligence-based automatic coffee quality sorting system.Keywords: Coffee Bean Classification; CNN-SVM; ResNet-50 AbstrakPenentuan mutu biji kopi di Indonesia umumnya masih dilakukan secara manual berdasarkan cacat fisik dan warna, yang bersifat subjektif dan memerlukan waktu lama. Penelitian ini bertujuan untuk mengembangkan model klasifikasi mutu biji kopi hijau berbasis citra digital menggunakan metode Convolutional Neural Network dan Support Vector Machine (CNN-SVM). CNN digunakan untuk ekstraksi fitur dengan arsitektur ResNet-50, sedangkan SVM berfungsi untuk klasifikasi menggunakan kernel Radial Basis Function (RBF). Dataset terdiri dari 10 kelas cacat biji kopi dan dibagi menjadi 80% data latih serta 20% data uji. Hasil pengujian menunjukkan nilai akurasi sebesar 77,68%, presisi 80,04%, recall 77,15%, dan f1-score 77,47%. Pendekatan ini membuktikan bahwa kombinasi CNN dan SVM mampu meningkatkan akurasi dan stabilitas model. Temuan ini menjadi kebaruan dalam pengembangan sistem sortasi mutu kopi otomatis yang efisien dan objektif berbasis kecerdasan buatan.Kata kunci: Klasifikasi Biji Kopi; CNN-SVM; ResNet-50
Co-Authors Gunawan, Michael Pratama, Samuel Effendi