Main Article Content
DDoS attack is a type of network security threat that aims to flood target networks with harmful traffic. Despite the fact that several statistical methods have been designed for DDoS attack detection, creating an offline or real-time detector with low cost is still one of the main concerns. The aim of this research using (ML) techniques is to categorize data traffic as either normal or malicious. This paper handles a general “PortScan attack CICDDOS2019 Data set” there are 79 attributes in total, and more than 200 thousand records, this dataset contains the normal and attack traffics. This study implements boruta, forward and backward, and variable significance algorithms using the RStudio tool to detect the most relevant attributes through selection feature and perform classification effectively. After the preprocessing and feature selection phases, the obtained dataset was classified by Random Forest (RF), Naïve Bayes (NB), and Support Vector Machine (SVM) algorithms. The experimental results show that (RF) with forward and backward has a higher rate of accuracy than other algorithms 100%, Precision 0.99993, Recall 100%, F1-Measure 0.99994, Specificity 0.9999 with classification achievement.