Hybrid Feature Selection Approach for Detecting Suspicious URLs in IoT Systems

Abstract

The rapid growth of Internet of Things (IoT) environments has significantly increased the risk of phishing attacks through malicious URLs, posing serious threats to data security and user privacy. This paper presents an efficient hybrid feature selection technique for the prediction of suspicious URLs using machine learning. The proposed system integrates a Gradient Boosting Classifier with a feature extraction mechanism to identify critical characteristics of URLs. A hybrid approach combining statistical and heuristic-based feature selection is employed to improve model accuracy while reducing computational complexity. The dataset is preprocessed and divided into training and testing subsets, enabling effective model evaluation. The developed web-based application, implemented using Flask, allows real-time URL analysis by extracting 30 distinct features and classifying them as safe or phishing. Experimental results demonstrate that the proposed method achieves high prediction accuracy and reliability, making it suitable for deployment in IoT-based security frameworks. The system enhances detection performance while maintaining efficiency, thereby contributing to robust cybersecurity solutions in resource-constrained IoT environments.