OptiPhish: URL-based phishing detection using an optimised feature framework

Phishing, a social engineering cyberattack, poses a significant risk to an individual's privacy and finances. To launch these attacks, criminals trick consumers into giving their sensitive information. Existing techniques lack precise and fast identification of phishing websites. This work suggests a method to build a strong model with high accuracy and low false positive rates for categorising URLs as benign or dangerous. We found optimised features via feature engineering, which helped to eliminate less significant ones. This approach reduced processing time and raised all evaluation criteria by providing optimised features to the machine learning classifier. A balanced dataset prevents underfitting and overfitting, ensuring a reasonable and adequate dataset for the machine learning systems. The experiments show that Random Forest and XGBoost algorithms perform best with 97% accuracy, a 0.8% false positive rate, and a 96.9% F1 score. This suggested work represents significant progress in enhancing cybersecurity.