A Comprehensive Analysis of Feature Selection and Classification Techniques for Web Attack Classification

Abstract

Cybersecurity is seriously threatened by web attacks, which makes it necessary to create strong classification models as soon as possible for early detection and prevention. Networked system security is now a crucial global concern affecting people, businesses, and governments. Attacks on networked systems are occurring far more often, and the attackers' strategies are always changing. One defense against these attacks is intrusion detection. Machine learning is a popular and useful method for creating intrusion detection systems (IDS). Having more representative and discriminative traits greatly enhances an IDS's performance. We do a thorough review of feature selection and classification methods for the purpose of classifying web attacks in this work. This work aims to use Decision Tree, Random Forest, and Logistic Regression as classifiers and evaluate their accuracy with the phishing website dataset. To improve the performance of the selected classifiers, subsequently, various feature selection methods, including Recursive Feature Elimination (RFE) and the Chi-square test, were used to identify the most relevant features for classification. Feature selection is a key topic for dimension reduction and classification in high-dimensional datasets. During the feature selection procedure, only the most relevant attributes from the datasets will be chosen. Logistic Regression, Decision Tree, and Random Forest. We assess the models' performance based on metrics such as accuracy and classification reports to determine their effectiveness in classifying web attacks. Our findings provide insights into the effectiveness of different feature selection and classification techniques for web attack classification, contributing to the advancement of cybersecurity research.