Implementasi Algoritma Naive Bayes Classifier (NBC) Dan Information Gain Untuk Mendeteksi DDoS

Abstract

In this study, feature selection was also carried out using the information gain method, the result of feature selection improve the performance of DDoS attack detection against the Naive Bayes Classifier classification algorithm. The results obtained in this study are system testing on the results of the comparison of data performance that has been selected using 17 features and without the application of information gain feature selection using 43 features of course different, there are superior results from the application of Information Gain feature selection with an average accuracy value of 75.81 %, while the results obtained without the application of feature selection are 75.57%. The average precision level system performance using 17 features is 91.61%, while average precision result using 43 features is 92.20%. For the average recall value using 17 features, it is 57.63%, and results recall uses 43 features by 57.31%. In terms of execution time, the time required to execute the program using 17 features is faster and more effective, namely 89.17 seconds, while the program execution time using 43 features is longer, namely 205.34 seconds.

Untitled

“…Dataset CSE-CIC-IDS2018 memiliki jumlah f itur sebanyak 80 f itur yang dimana jumlah data serangan DoS yaitu sebanyak 601.802 data dari total keseluruhan data sebanyak 1.048.574. pada tabel di bawah ini merupakan rincian dari label data pada dataset.Tabel 1. Rincian Data Target untuk membantu penyerangan terhadap komputer target yang menyebabkan komputer target menjadi lambat hingga sistem komputer menjadi error[5]. Ada berbagai tipe DoS salah satunya yaitu Application Layer Denial of Service.…”