An Email Spam Filtering Model Using Ensemble of Machine Learning Techniques

Abstract

The growth of spam emails is on the increase responsible for larger portions of the global email traffics. Aside the annoyance and the time wasted sifting through the unwanted messages; spam emails can also cause immeasurable harms through malicious software capable of damaging systems and compromising confidential information. The risks of filtering spam emails is that sometimes, legitimate mails are marked as spam, yet the results of not filtering spam are the constant flood of spam clogs on networks that adversely impacts users inboxes while draining valuable resources on the networks such as bandwidth and storage capacity, productivity loss and interfere with the expedient delivery of legitimate emails. Several researchers had worked on the design of models for spam email filtering using different techniques, however the detection accuracy of these models have also become subject of discussions. This study developed spam email filtering model using Ensemble of Decision Tree, Support Vector Machine and Multilayer Perceptron (DT-SVM-MLP) technique as a solution approach to solving issues of low spam emails detection accuracy. The ensemble model was trained using forward propagation training technique and the performance was evaluated using five performance metrics of Accuracy, False Positive (FP) Rate, Precision, Recall and F-Measure.