KLASIFIKASI SERANGAN DDOS UDP FLOOD MENGGUNAKAN KNN++ DENGAN ANOVA FEATURE SELECTION

Satrio, Muhammad Bagus (2025) KLASIFIKASI SERANGAN DDOS UDP FLOOD MENGGUNAKAN KNN++ DENGAN ANOVA FEATURE SELECTION. Undergraduate thesis, Universitas Pembangunan Nasional Veteran Jawa Timur.

	Text (Cover) 21081010102_Cover.pdf Download (3MB)
	Text (Bab I) 21081010102_Bab 1.pdf Download (501kB)
	Text (Bab II) 21081010102_Bab 2.pdf Restricted to Registered users only until 12 June 2028. Download (996kB) | Request a copy
	Text (Bab III) 21081010102_Bab 3.pdf Restricted to Registered users only until 12 June 2028. Download (1MB) | Request a copy
	Text (Bab IV) 21081010102_Bab 4.pdf Restricted to Registered users only until 12 June 2028. Download (4MB) | Request a copy
	Text (Bab V) 21081010102_Bab 5.pdf Restricted to Registered users only until 12 June 2028. Download (399kB) | Request a copy
	Text (Daftar Pustaka) 21081010102_Daftar Pustaka.pdf Download (502kB)

Abstract

Detection of Distributed Denial of Service (DDoS) attacks is a major challenge in modern network security. This research proposes a DDoS attack classification approach using modified K-Nearest Neighbor algorithms, namely KNN++ and CD-KNN, with the utilization of ANOVA feature selection as well as SMOTE oversampling method to handle data imbalance. The CIC-DDoS2019 dataset is used as the main data source which has been preprocessed and normalized. In testing without SMOTE, the combination of KNN++ with ANOVA feature selection (α = 0.05) produces the highest accuracy of 99.994% and the ROC-AUC value reaches 99.54% at parameter K = 3 with a computation time of 1946 seconds, then for the combination of CD-KNN with ANOVA reaches an accuracy of 99.993% with a ROC-AUC value of 2055 seconds showing excellent and efficient classification performance. While in testing with SMOTE, KNN++ achieves the same high performance at parameter K=3 with an accuracy of 99.994% and ROC-AUC reaching 99.76%, but accompanied by a significant increase in computation time of 3846 seconds while for CD-KNN with Alpha=6 achieves the same highest accuracy results as KNN++ which is 99.994% with an ROC-AUC value of 99.76% with a computation time of 3903 seconds. The results show that SMOTE is able to improve the stability of the evaluation metric on unbalanced data, while ANOVA feature selection consistently improves the efficiency and accuracy of the model. Compared to CD-KNN, the KNN++ algorithm shows more consistent and efficient performance in various configurations.

Item Type:	Thesis (Undergraduate)
Contributors:	Contribution Contributors NIDN/NIDK Email Thesis advisor Anggraeny, Fetty Tri NIDN0711028201 fettyanggraeny.if@upnjatim.ac.id Thesis advisor Junaidi, Achmad NIDN0710117803 achmadjunaidi.if@upnjatim.ac.id
Subjects:	T Technology > T Technology (General)
Divisions:	Faculty of Computer Science > Departemen of Informatics
Depositing User:	Muhammad Bagus Satrio
Date Deposited:	12 Jun 2025 08:04
Last Modified:	12 Jun 2025 08:04
URI:	https://repository.upnjatim.ac.id/id/eprint/37418

Actions (login required)

View Item