IMPLEMENTASI ALGORITMA RANDOM FOREST DALAM KLASIFIKASI SERANGAN DISTRIBUTED DENIAL OF SERVICE MENGGUNAKAN KOMBINASI SELEKSI FITUR

Fahriza Ahmad Harits, - (2025) IMPLEMENTASI ALGORITMA RANDOM FOREST DALAM KLASIFIKASI SERANGAN DISTRIBUTED DENIAL OF SERVICE MENGGUNAKAN KOMBINASI SELEKSI FITUR. S1 thesis, Universitas Pendidikan Indonesia.

Abstract

Serangan Distributed Denial of Service (DDoS) merupakan ancaman siber yang terus meningkat dan mengancam ketersediaan layanan digital. Meskipun machine learning, khususnya algoritma Random Forest (RF), menunjukkan efektivitas tinggi dalam deteksi serangan, efisiensi komputasi dan akurasi model masih dapat dioptimalkan melalui seleksi fitur yang cermat. Penelitian ini bertujuan untuk mengembangkan dan mengevaluasi model deteksi DDoS menggunakan algoritma Random Forest yang dioptimalkan dengan kombinasi metode seleksi fitur Mutual Information (MI) dan Recursive Feature Elimination with Cross-Validation (RFECV). Model dikembangkan menggunakan dataset publik CIC-DDoS2019. Proses seleksi fitur dilakukan dalam dua tahap: MI untuk menyaring fitur informatif awal, diikuti oleh RFECV untuk mendapatkan subset fitur optimal. Performa model dievaluasi berdasarkan metrik akurasi, presisi, recall, F1-score, ROC-AUC, serta efisiensi komputasi yang diukur dari waktu pelatihan, waktu deteksi, dan penggunaan memori. Hasil menunjukkan bahwa kombinasi MI-RFECV berhasil mereduksi 85 fitur menjadi 28 fitur optimal. Model dengan fitur terseleksi ini mencapai performa klasifikasi yang sangat tinggi, dengan akurasi 100%, presisi 99,99%, recall 100%, F1-score 100%, dan ROC-AUC 100%. Selain itu, model menunjukkan efisiensi signifikan dengan waktu pelatihan hanya 15,62 detik dan penggunaan memori 17,27 MiB. Kombinasi seleksi fitur MI-RFECV terbukti tidak hanya mampu menyederhanakan model dan meningkatkan efisiensi komputasi secara drastis, tetapi juga mempertahankan bahkan memperkuat akurasi klasifikasi serangan DDoS oleh algoritma Random Forest. ---------- Distributed Denial of Service (DDoS) attacks constitute an escalating cyber threat that jeopardizes the availability of digital services. While machine learning, particularly the Random Forest (RF) algorithm, demonstrates high efficacy in attack detection, its computational efficiency and model accuracy can be further optimized through meticulous feature selection. This study aims to develop and evaluate a DDoS detection model by optimizing the Random Forest algorithm with a combined Mutual Information (MI) and Recursive Feature Elimination with Cross-Validation (RFECV) feature selection approach. The model was trained and tested on the public CIC-DDoS2019 dataset. The feature selection process was executed in two stages: MI was first applied to filter for informative features, followed by RFECV to identify the most optimal feature subset. Model performance was evaluated using accuracy, precision, recall, F1-score, and ROC-AUC metrics, alongside computational efficiency, which was assessed by measuring training time, detection time, and memory consumption. The results demonstrate that the combined MI-RFECV approach successfully reduced the feature set from 85 to an optimal 28. The resulting model achieved exceptional classification performance, yielding an accuracy of 100%, precision of 99.99%, recall of 100%, F1-score of 100%, and ROC-AUC of 100%. Furthermore, the model demonstrated significant efficiency, with a training time of just 15.62 seconds and a memory consumption of 17.27 MiB. The combined MI-RFECV feature selection method is proven to not only simplify the model and drastically improve computational efficiency but also to maintain and even enhance the classification accuracy of the Random Forest algorithm for DDoS attacks.

	Text S_RPL_2100185_Title.pdf Download (2MB)
	Text S_RPL_2100185_Chapter1.pdf Download (405kB)
	Text S_RPL_2100185_Chapter2.pdf Restricted to Staf Perpustakaan Download (991kB) | Request a copy
	Text S_RPL_2100185_Chapter3.pdf Download (760kB)
	Text S_RPL_2100185_Chapter4.pdf Restricted to Staf Perpustakaan Download (934kB) | Request a copy
	Text S_RPL_2100185_Chapter5.pdf Download (287kB)
	Text S_RPL_2100185_Appendix.pdf Restricted to Staf Perpustakaan Download (645kB) | Request a copy

Item Type:	Thesis (S1)
Additional Information:	https://scholar.google.com/citations?hl=id&user=Y2YOH60AAAAJ&scilu=&scisig=ACUpqDcAAAAAaJTehoZHBIh3ldG9HX82bZRMX_U&gmla=AH8HC4yiybAZPO9m5zlNCIeCDxRSluivgna4k9B5tNqHyOpOOKR1imWpYz96pAUYqtxA6mo26XUfUPvwaLQRiR5MGW6M2SUUr1EfL448cNBLOkTihUvVeI12A7A&sciund=13798478718846049048 ID Sinta Dosen: Raditya Muhammad: 6682222 Mochamad Iqbal Ardimansyah: 6658552
Uncontrolled Keywords:	Keamanan Jaringan, Deteksi DDoS, Random Forest, Seleksi Fitur, Mutual Information, Recursive Feature Elimination Cross-Validation, Network Security, DDoS Detection, Random Forest, Feature Selection, Mutual Information, Recursive Feature Elimination with Cross-Validation.
Subjects:	L Education > L Education (General) Q Science > QA Mathematics > QA75 Electronic computers. Computer science Q Science > QA Mathematics > QA76 Computer software T Technology > T Technology (General)
Divisions:	UPI Kampus cibiru > S1 Rekayasa Perangkaat Lunak
Depositing User:	Fahriza Ahmad Harits
Date Deposited:	19 Aug 2025 07:09
Last Modified:	19 Aug 2025 07:09
URI:	http://repository.upi.edu/id/eprint/135322

Actions (login required)

View Item