<![CDATA[Nowadays, email is a communication media that is often used in the digital era, with various advantages offered by email, accompanied by the rise of email spam which can disrupt the comfort of its users and accessibility on the email service provider platform. Using manual spam filtering techniques has proven to be very time-consuming and labor-intensive, so an alternative technique is needed that can perform sorting automatically using Machine Learning. This research aims to develop a form of spam detection model that uses a mixed Naive Bayes approach that combines various forms of TF-IDF feature representation with various statistical features that can calculate message length, number of capital letters, and various number of links, and compare its performance with various other algorithm approaches consisting of Support Vector Machine, Logistic Regression, and Random Forest, this study uses a public dataset containing examples of 5,572 emails containing important emails and spam emails combined. The evaluation form will be calculated using the metrics Accuracy, Precision, Recall, F1-Score, and Training Time. The results of the experiment explain that Naive Bayes with Mixture is able to produce an accuracy of 96.4% with advantages in calculating computational efficiency, but Random Forest has the highest accuracy level reaching 97.9%. So it shows that this research proves that Naive Bayes with various mixed approaches is worthy of being applied to an Email Spam detection system that requires high speed and efficiency.]]>
Copyrights © 2025
