Evaluation of Machine Learning Techniques for Email Spam Classification

Electronic mail (Email) is one of the official and very common way of exchanging data and information over digital and electronic devices. Millions of users worldwide use email to exchange data and information between email servers. On the other hand, unwanted emails or spam became phenomenon challenging major companies and organizations due to the volume of spam which is increasing dramatically every year. Spam is annoying and may contain harmful contents. In addition, spam consume computers, servers, and network resources, causes harmful bottleneck, effect on computing memory and speed of digital devices. Moreover, the time consumed by the users to remove unwanted emails is huge. There are many methods developed to filter spam like keyword matching blacklist/whitelist and header information processing. Though, classical methods like blocking the source to prevent the spam are not effective. This study demonstrates and reviews the performance evaluation of the most popular and effective machine learning techniques and algorithms such as Support Vector Machine, ANN, J48, and Naïve Bayes for email spam classification and filtering. In con conclusion, support vector machine performs better than any individual algorithm in term of accuracy. This research contributes on the for the development of methods and techniques for better detection and prevention of spam.