A Comparative Study of Two Different Spam Detection Methods

Abstract

With the development of the Internet, the problem of spam has become more and more prominent. Attackers can spread viruses through spam or place malicious advertisements, which have seriously interfered with people’s life and internet security. Therefore, it is of great significance to study efficient spam detection methods. Currently using machine learning methods for spam detection has become a mainstream direction. In this paper, the machine learning method of Bayesian linear regression and decision forest regression are used to conduct experiments on a data set from UCI Machine Learning Repository. We use the trained models to predict whether a mail is spam or not, and find better prediction scheme by comparing quantitative results. The experimental results show that the method of decision forest regression can get better performance and is suitable for numerical prediction.