Malicious Website Detection using Machine Learning
Sanjay Kumar, K Sri Krishna, Santhosh Jerome AR, Vimanthann S. Barath · 2025
With the recent rise in internet usage on a global level, there is an ever-increasing threat to user privacy and software security. Fake and unsafe websites are often used to manipulate unaware users into giving their private credentials to individuals or organizations with malicious intentions. In order to improve cybersecurity, this study suggests a machine learning-based method for detecting dangerous websites. The process involves collecting data from labelled URLs and then extracting features based on important attributes including IP presence, URL length, and special characters. In order to preprocess the extracted features, categorical attributes are transformed into numerical form. Support Vector Machine (SVM), Random Forest, and XGBoost are among the machine learning models that are taught to categorize URLs as either safe or dangerous in this current research. The top-performing model is chosen for real-time deployment after the models are assessed using accuracy, precision, recall, and F1-score. Results from experiments show how well each of the suggested method works to detect dangerous websites with a high degree of accuracy, providing a flexible defense against online threats. Random Forest Classifier had the highest accuracy value of 98.17% out of all the models used to detect malicious websites.