Machine learning algorithm to identifies fraud emails with feature selection
Anita Sindar Sinaga, Musthafa Haris Munandar, Arjon Samuel Sitio · IOP Conference Series Materials Science and Engineering · 2021
Abstract One percent of the emails that come in each day are fraudulent. Promotion emails tend to offer products. The recipient’s email is recorded by a company or organization. Not all promotional emails are considered spam or hoaxes. When observed, incoming promotional emails provide the information that is needed. How to identify promotional emails including hoaxes or shipping with machine learning, known as algorithms, Support Vector Machines, Naïve Bayes, Decision Tree, Logistic Regression, Stochastic Gradient Descent, and Neural Network (MLP). Decision trees are effective tracing using a data structure consisting of vertices & edges. A node (root, branch, leaf) can categorize incoming e-mail, including hoax or e-mail shipping. Previously, it was necessary to categorize the characteristics of the email. After searching the email, the promotional email grouping is continued. In Feature Extraction, the calculation of Gain and Entropy is used to determine the selection of features in the classification of promotional emails, fraudulent emails or hoaxes.