E-Commerce Fraud Detection Using Generated Data From BANKSIM Using Machine Learning Approach: A Pilot Study

E-commerce has become a new habit to meet daily needs. The rapid development of technology makes e-commerce activities easy to carry out and almost anyone can do it. Unfortunately, these activities not only bring profits, but e-commerce activities also involve losses and undesirable actions. One of the losses in e-commerce operations is fraud in making transactions. To prevent fraud, a special method is needed that can detect whether the transaction performed is fraudulent or not. One of these methods is fraud detection, which can be done with a machine learning approach. This pilot study focuses on solving the problem of fraudulent transactions by creating a machine learning model with the Gaussian Naïve Bayes, K-NN, and Fine Tree algorithms. The dataset used in this study comes from atapdata.ai, which is expected to create a fraudulent transaction detection model with good accuracy. This pilot study shows that Fine Tree has the highest accuracy value with an accuracy value of 99.5% and an F1-Score of 0.997851.