PERBANDINGAN KINERJA ALGORITMA RANDOM FOREST DAN LIGHT GRADIENT BOOSTING MACHINE UNTUK DETEKSI DOMAIN PHISING BERBASIS ANALISIS TYPOSQUATTING

Authors

  • Puteri Elisabeth Pakpahan Teknik Informatika, Fakultas Teknik, Universitas Krisnadwipayana
  • Nuke Lu’lu Ul Chusna Teknik Informatika, Fakultas Teknik, Universitas Krisnadwipayana
  • Harry Dwiyana Kartika Teknik Informatika, Fakultas Teknik, Universitas Krisnadwipayana

DOI:

https://doi.org/10.35316/jimi.v11i2.10844

Keywords:

Phishing;, Typosquatting; , Random Forest; , LightGBM;, SHAP;

Abstract

Typosquatting-based phishing registers domains that resemble official brands to deceive users, and blacklist-based protection cannot stop newly registered look-alike domains. This study compares Random Forest and Light Gradient Boosting Machine (LightGBM) for phishing domain detection using typosquatting, lexical, suspicious-keyword, and top-level-domain features, and deploys the best model in a Streamlit simulation with SHAP interpretability. Following CRISP-DM, the dataset contains 7,999 domains after cleaning (5,000 legitimate domains from the Tranco list and 2,999 phishing domains generated for 13 target brands through the HaveIBeenSquatted API), described by 10 numeric features. Class imbalance was handled with class_weight='balanced', and both models were tuned with GridSearchCV on a stratified 70:30 split. With phishing-class recall as the main criterion, the tuned LightGBM reached an accuracy of 94.37%, a phishing precision of 95.16%, and a phishing recall of 89.56%; the tuned Random Forest reached 94.46%, 95.71%, and 89.22%. The difference is small, and LightGBM was selected for its higher recall. A local SHAP explanation showed jaro_winkler_score as the largest contributor for a sample domain. The prototype reports a prediction, a confidence score, and a feature-contribution plot.

Downloads

Download data is not yet available.

References

[1] Badan Siber dan Sandi Negara, “Lanskap keamanan siber Indonesia,” Id-SIRTII/CC, vol. 70, pp. 1–107, 2024.

[2] Asosiasi Penyelenggara Jasa Internet Indonesia, “Profil Internet Indonesia 2026,” APJII, Jakarta, 2026.

[3] A. F. Mahmud and S. Wirawan, “Deteksi phishing website menggunakan machine learning metode klasifikasi,” Sistemasi: Jurnal Sistem Informasi, vol. 13, no. 4, 2024.

[4] Lukito and W. B. T. Handaya, “Deteksi website phishing menggunakan teknik machine learning,” Jurnal Informatika Atma Jogja, vol. 6, no. 1, pp. 69–80, 2025, doi: 10.24002/jiaj.v6i1.11538.

[5] Anti-Phishing Working Group, “Phishing activity trends report: 1st quarter 2025,” 2025. [Online]. Available: https://docs.apwg.org/reports/apwg_trends_report_q1_2025.pdf

[6] Y. Y. Santika, R. Rianto, and E. Ujianto, “Studi komprehensif keamanan siber: perbandingan teknologi AI dengan sistem non-AI dalam deteksi dan pencegahan ancaman,” Jurnal Komtika (Komputasi dan Informatika), vol. 9, no. 1, pp. 45–64, 2025, doi: 10.31603/komtika.v9i1.13149.

[7] A. M. Almuhaideb et al., “Homoglyph attack detection model using machine learning and hash function,” 2022.

[8] S. Agarwal and M. Vasek, “Examining newly registered phishing domains at scale,” Journal of Cybersecurity, vol. 12, doi: 10.1093/cybsec/tyag020.

[9] R. Fauzan, A. V. Vitianingsih, D. Cahyono, A. L. Maukar, and Y. A. B. Suprio, “Application of classification algorithms in machine learning for phishing detection,” Indonesian Journal of Machine Learning and Computer Science, vol. 5, no. 2, pp. 531–540, 2025.

[10] N. B. Putri and A. W. Wijayanto, “Analisis komparasi algoritma klasifikasi data mining dalam klasifikasi website phishing,” Komputika: Jurnal Sistem Komputer, vol. 11, no. 1, pp. 59–66, 2022, doi: 10.34010/komputika.v11i1.4350.

[11] D. B. Suwarno and M. Hardjianto, “Deteksi website phishing dari analisis URL menggunakan algoritma Random Forest,” vol. 21, no. 2, pp. 145–152, 2024.

[12] P. R. E. Gabriel, “Deteksi URL phishing menggunakan hybrid deep learning CNN dan XGBoost dengan teknik balancing SMOTE-ENN,” vol. 5, no. 2, pp. 70–79, 2025.

[13] Y. Perdana, “Comparative analysis of Random Forest and XGBoost for detecting phishing websites: a machine learning approach,” vol. 7, no. 2, pp. 906–921, 2025.

[14] C. Feresa, M. Foozy, M. Amir, I. Anuar, A. Maslan, and H. A. Mohd, “Phishing URLs detection using Naïve Bayes, Random Forest and LightGBM algorithms,” vol. 5, no. 1, pp. 56–63, 2024.

[15] P. S. Rizky, R. H. Hirzi, and U. Hidayaturrohman, “Perbandingan metode LightGBM dan XGBoost dalam menangani data dengan kelas tidak seimbang,” 2022.

[16] M. C. Calzarossa and R. Zieni, “Explainable machine learning for phishing feature detection,” pp. 362–373, 2024, doi: 10.1002/qre.3411.

[17] M. Sipper, “High per parameter: a large-scale study of hyperparameter tuning for machine learning algorithms,” pp. 1–13, 2022.

[18] O. Rainio, J. Teuho, and R. Klén, “Evaluation metrics and statistical tests for machine learning,” Scientific Reports, pp. 1–14, 2024, doi: 10.1038/s41598-024-56706-x.

[19] S. Remya, M. J. Pillai, K. K. Nair, S. R. Subbareddy, and Y. Y. Cho, “An effective detection approach for phishing URL using ResMLP,” IEEE Access, vol. 12, pp. 79367–79382, 2024, doi: 10.1109/ACCESS.2024.3409049.

[20] E.-S. M. El-kenawy, M. M. Eid, H. L. Hussein, A. M. Osman, and A. M. Elshewey, “Optimized deep learning model using binary particle swarm optimization for phishing attack detection: a comparative study,” vol. 5, no. 2, pp. 685–702, 2025.

[21] R. M. Aziz, R. Mahto, K. Goel, A. Das, and P. Kumar, “Modified genetic algorithm with deep learning for fraud transactions of Ethereum smart contract,” Applied Sciences, 2023.

Published

16-11-2026

How to Cite

Pakpahan, P. E., Chusna, N. L. U., & Kartika, H. D. (2026). PERBANDINGAN KINERJA ALGORITMA RANDOM FOREST DAN LIGHT GRADIENT BOOSTING MACHINE UNTUK DETEKSI DOMAIN PHISING BERBASIS ANALISIS TYPOSQUATTING. Jurnal Ilmiah Informatika, 11(2), 221–229. https://doi.org/10.35316/jimi.v11i2.10844

Similar Articles

You may also start an advanced similarity search for this article.