Machine Learning based Spam E-Mail Detection

Spam email is one of the biggest issues in the world of internet. Spam emails not only influence the organisations financially but also exasperate the individual email user. This paper aims to propose a machine learning based hybrid bagging approach by implementing the two machine learning algorithms: Nave Bayes and J48 (decision tree) for the spam email detection. In this process, dataset is divided into different sets and given as input to each algorithm. Total three experiments are performed and the results obtained are compared in terms of precision, recall, accuracy, f-measure, true negative rate, false positive rate and false negative rate. The two experiments are performed using individual Nave Bayes & J48 algorithms. Third experiment is the proposed SMD system implemented using hybrid bagged approach. The overall accuracy of 87.5% achieved by the hybrid bagged approach based SMD system.