Detecting Deceptive URLs: Exploring Logistic Regression and Decision Tree Models

This research investigates the detection of phishing websites utilizing machine learning techniques and feature selection methodologies. Phishing remains a prevalent cyber threat, posing significant risks to users' sensitive information and online security. This study proposes a comprehensive approach that combines Chi-squared and XGBoost feature selection methods with logistic regression and Decision Tree classification models. By leveraging these techniques, it aims to achieve accurate and reliable identification of phishing websites while maintaining interpretability and generalizability. The findings underscore the importance of strategic feature selection and model building in enhancing the effectiveness of phishing detection systems. The results demonstrate promising outcomes, with both logistic regression and Decision Tree models exhibiting high accuracy in distinguishing between legitimate and phishing websites. Additionally, the key challenges such as feature scaling have been addressed to mitigate potential biases and data leakage. This research contributes to the advancement of cybersecurity measures, offering insights and methodologies to combat the persistent threat of phishing in the digital landscape.