Analysis of Email Spam Detection Using Naive Bayes and Support Vector Machine Classification Algorithms

This study focused on analyzing email spam detection using two classification algorithms. The research was conducted using a dataset of 5,172 emails. The goal was to compare the effectiveness of these two machine learning techniques for classifying emails. The analysis involved evaluating the performance of SVM and a novel Naive Bayes Classifier. For the experiment, a sample size of 25 emails per group was determined using a G power analysis with 80% power. The results showed that while SVM achieved an accuracy of 89.78% with a relatively low mean error, the novel Naive Bayes Classifier outperformed it with an accuracy of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 4. 0 9 \%}$</tex>. This indicates that the Naive Bayes Classifier is more effective at accurately classifying emails as spam or not spam compared to SVM. The statistical significance of the results was confirmed with a p-value of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$0.036(p&lt;0.05)$</tex>, highlighting that the Naive Bayes Classifier's superior accuracy is statistically significant. In summary, the study demonstrates that the Naive Bayes Classifier provides better performance in email spam detection than the Support Vector Machine.