Perbandingan Metode Supervised Machine Learning – Support Vector Machine (SVM) dan K-Nearest Neighbor Classifier (KNN) dalam Analisis Pola Cyber Attack Pada Dataset Network Intrusion Detection System (NIDS)

Shiddiq, Haikal (2024) Perbandingan Metode Supervised Machine Learning – Support Vector Machine (SVM) dan K-Nearest Neighbor Classifier (KNN) dalam Analisis Pola Cyber Attack Pada Dataset Network Intrusion Detection System (NIDS). S2 thesis, Universitas Mercu Buana - Menteng.

	Text (COVER) 55421110004-Haikal Shiddiq-01 Cover - Haikal Shiddiq.pdf Download (646kB)
	Text (BAB I) 55421110004-Haikal Shiddiq-02 Bab 1 - Haikal Shiddiq.pdf Restricted to Registered users only Download (86kB)
	Text (BAB II) 55421110004-Haikal Shiddiq-03 Bab 2 - Haikal Shiddiq.pdf Restricted to Registered users only Download (319kB)
	Text (BAB III) 55421110004-Haikal Shiddiq-04 Bab 3 - Haikal Shiddiq.pdf Restricted to Registered users only Download (430kB)
	Text (BAB IV) 55421110004-Haikal Shiddiq-05 Bab 4 - Haikal Shiddiq.pdf Restricted to Registered users only Download (1MB)
	Text (BAB V) 55421110004-Haikal Shiddiq-06 Bab 5 - Haikal Shiddiq.pdf Restricted to Registered users only Download (44kB)
	Text (DAFATR PUSTAKA) 55421110004-Haikal Shiddiq-08 Daftar Pustaka - Haikal Shiddiq.pdf Restricted to Registered users only Download (143kB)
	Text (LAMPIRAN) 55421110004-Haikal Shiddiq-09 Lampiran - Haikal Shiddiq.pdf Restricted to Registered users only Download (415kB)

Abstract

Penelitian ini bertujuan untuk menganalisis dataset opensource NF-UQ-NIDS-v2 menggunakan supervised learning - Support Vector Machine (SVM) dan K-Nearest Neighbor Classifier (KNN) dalam menghitung tingkat akurasi, precision, recall dan f1-score pada satuan persentase. Metode SVM dan KNN digunakan dengan training dan test menggunakan cross validation (cv) sebanyak 10 kali dan dataset tersebut dibagi menjadi 4 bagian besar agar proses analisisnya lebih mudah. Hasil menunjukkan metode K-Nearest Neighbor (KNN) memiliki akurasi lebih tinggi yaitu >90% dibandingkan metode Support Vector Machine (SVM) untuk seluruh grup dataset. Namun untuk membuat sebuah prediksi terkait jumlah tipe serangan yang sama cenderung menggunakan model Support Vector Machine (SVM) yang terlihat dari heatmap prediction. Pola serangan DoS/DDoS paling sering terjadi dan metode Support Vector Machine (SVM) serta K-Nearest Neighbor Classifier (KNN) efektif untuk menganalisis pola serangan. Data ini bersifat opensource sehingga hasil training dan test dapat berubah dengan update data. Dengan pola serangan DoS/DDoS ini diharapkan setiap pemilik sistem aplikasi berbasis web server yang di publish ke internet dapat melakukan pencegahan serangan tersebut dengan memperkuat keamanan akses di jaringan dan sistem This study aims to analyze the opensource NF-UQ-NIDS-v2 dataset using supervised learning - Support Vector Machine (SVM) and K-Nearest Neighbor Classifier (KNN) in calculating the level of accuracy, precision, recall and f1-score in percentage units. The SVM and KNN methods are used with training and testing using cross validation (cv) 10 times and the dataset is divided into 4 large parts to make the analysis process easier. The results show that the K-Nearest Neighbor (KNN) method has a higher accuracy of >90% compared to the Support Vector Machine (SVM) method for all dataset groups. However, to make a prediction regarding the number of the same type of attack, the Support Vector Machine (SVM) model tends to be used as seen from the heatmap prediction. The DoS/DDoS attack pattern occurs most often and the Support Vector Machine (SVM) and K-Nearest Neighbor Classifier (KNN) methods are effective in analyzing attack patterns. This data is opensource so that the training and test results can change with data updates. With this DoS/DDoS attack pattern, it is hoped that every owner of a web server-based application system published to the internet can prevent such attacks by strengthening access security on the network and system.

Item Type:	Thesis (S2)
NIM/NIDN Creators:	55421110004
Uncontrolled Keywords:	Supervised learning, Support Vector Machine (SVM), K-Nearest Neighbor Classifier (KNN), dataset, cyber security, data analysis, analisis serangan cyber Supervised learning, Support Vector Machine (SVM), K-Nearest Neighbor Classifier (KNN), dataset, cyber security, data analysis, cyber attack analysis
Subjects:	600 Technology/Teknologi > 620 Engineering and Applied Operations/Ilmu Teknik dan operasi Terapan > 621 Applied Physics/Fisika terapan
Divisions:	Pascasarjana > Magister Teknik Elektro
Depositing User:	SILMI KAFFA MARISKA
Date Deposited:	20 Sep 2024 03:58
Last Modified:	20 Sep 2024 03:58
URI:	http://repository.mercubuana.ac.id/id/eprint/91608

Actions (login required)

View Item