Standard Scaling and PCA-based Logistic Regression Model for Classifying and Detecting DDoS Attack

Abstract

Distributed Denial of Service Attack (DDoS) is one of the attacks used by cybercriminals against the target system in the world of technology. Using this attack, attackers tried to fool the target system and force the targeted system to deny the request of genuine clients. As a result of a successful attack, the targeted system becomes unavailable from the Internet, and the client gets a message of service unavailability. To perform this attack, the attacker tries to create fake traffic by using different approaches. This fake traffic forces the targeted system to deny the request of further requests received by genuine clients. Here DDoS stands for “Denial of Service,” which means denying clients' requests. For securing the services from DDoS attacks, here we propose a machine learning model for detecting and classifying DDoS attacks. This will help us to safeguard the services from this type of attack. In this paper, we are going to propose two different machine learning models which will help us to classify and detect the DDoS attack performed on the system. For creating the first model, we used standard scaling to pre-process the data. For the second model, we apply Principal Component Analysis(PCA) over the standard scaling to reduce the features present in the dataset so that the complexity of the model can be reduced. After preprocessing the dataset, we apply Logistic Regression in the dataset and build two different models, which help us to classify and detect the DDoS Attack.