Significance of Machine Learning for Detection of Malicious Websites on an Unbalanced Dataset

It is hard to trust any data entry on online websites as some may be malicious, and gather for illegal or unintended use. For example, bank login credit card information can misused financial theft. To make users aware of the digital safety websites, we have tried identify learn pattern a dataset consisting features malicious benign websites. We treated problem differentiation between classification applied several machine learning techniques, random forest, decision tree, logistic regression, support vector machines this data. Several evaluation metrics such accuracy, precision, recall, F1 score, false positive rate, were used evaluate performance each technique. Since was imbalanced, models developed bias during training toward specific class Multiple balancing undersampling, oversampling, SMOTE, removing bias. Our experiments showed that after data, forest algorithm using oversampling technique best results in all website feature dataset.