Implementasi Smote Pada Sistem Deteksi Intrusi Berbasis Machine Learning

Widodo, Akdeas Oktanae (2023) Implementasi Smote Pada Sistem Deteksi Intrusi Berbasis Machine Learning. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05211940000066_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 September 2025. Download (2MB) | Request a copy

Abstract

Sistem deteksi intrusi jaringan merupakan salah satu teknologi keamanan jaringan yang efisien untuk mendeteksi aktivitas berbahaya dalam sistem TI tradisional. Namun, masalah utama pada sistem ini adalah sering terjadi kesalahan dalam mengklasifikasikan aktivitas normal sebagai serangan atau false negative (FN) karena dataset yang digunakan tidak seimbang. Penelitian ini mengusulkan sistem deteksi intrusi berbasis machine learning menggunakan metode Synthetic Minority Over-sampling Technique (SMOTE) untuk mengatasi masalah ketidakseimbangan dataset. Penelitian ini menggunakan dataset NSL-KDD dan CIC-IDS2017 yang memiliki ketidakseimbangan data. Praproses data dilakukan dengan pembersihan data, eksplorasi data, normalisasi data, dan seleksi fitur menggunakan Recursive Feature Elimination (RFE). Metode (SMOTE) diterapkan untuk mengatasi masalah dataset yang tidak seimbang. Beberapa metode klasifikasi machine learning digunakan untuk menganalisis model terbaik yang paling optimal dalam sistem deteksi intrusi, antara lain Logistic Regression, Support Vector Machine (SVM), Naïve Bayes, Decision Tree, Random Forest, dan XGBoost. Model terbaik dipilih dan dilakukan pemetaan performa menggunakan confussion matrix. Hasil terbaik yang didapatkan setelah penerapan SMOTE untuk NSL-KDD dengan model Random Forest yaitu accuracy 0,9956, avg. precision 0,9884, avg. recall 0,9773, dan avg. f1-score 0,9827. Lalu, untuk model XGBoost yaitu accuracy 0,9964, avg. precision 0,9865, avg. recall 0,9812, dan avg. f1-score 0,9838. Sedangkan, untuk dataset CIC-IDS2017 dengan model Random Forest yaitu accuracy 0,9981, avg. precision 0,9458, avg. recall 0,8933, dan avg. f1-score 0,9147. Lalu, untuk model XGBoost yaitu accuracy 0,9989, avg. precision 0,9589, avg. recall 0,9385, dan avg. f1-score 0,9478.
=======================================================================================================================================
Intrusion detection system (IDS) is an efficient network security technology for detecting malicious activities in traditional IT systems. However, the main problem with this system is the frequent misclassification of normal activities as attacks or false negatives (FN) due to an imbalanced dataset. This study uses NSL-KDD and CIC-IDS2017 datasets to overcome this issue. Data preprocessing involves data cleaning, data exploration, data normalization, and feature selection using Recursive Feature Elimination (RFE). The Synthetic Minority Over-sampling Technique (SMOTE) method is applied to address the imbalanced dataset problem. Several machine learning classification methods are used to analyze the optimal model in intrusion detection systems, including Logistic Regression, Support Vector Machine (SVM), Naïve Bayes, Decision Tree, Random Forest, and XGBoost. The best model is selected and performance mapping is performed using confusion matrix. The best results obtained after SMOTE implementation for NSL-KDD with the Random Forest model were accuracy 0.9956, avg. precision 0.9884, avg. recall 0.9773, and avg. f1-score 0.9827. Meanwhile, for the XGBoost model, the results were accuracy 0.9964, avg. precision 0.9865, avg. recall 0.9812, and avg. f1-score 0.9838. For the CIC-IDS2017 dataset, the results for the Random Forest model were accuracy 0.9981, avg. precision 0.9458, avg. recall 0.8933, and avg. f1-score 0.9147. Meanwhile, for the XGBoost model, the results were accuracy 0.9989, avg. precision 0.9589, avg. recall 0.9385, and avg. f1-score 0.9478.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Deteksi Intrusi, SMOTE, Machine Learning, Imbalanced Dataset, Intrusion Detection.
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Akdeas Oktanae Widodo
Date Deposited:	21 Jul 2023 03:10
Last Modified:	21 Jul 2023 03:10
URI:	http://repository.its.ac.id/id/eprint/98834

Actions (login required)

View Item