Data Balance Optimization of Fraud Classification for E-Commerce Transaction
Aida Fitriyani, Wowon Priatna, Tyastuti Sri Lestari, Dwipa Handayani, Tb Ai Munandar, Amri Amri · 2022 Seventh International Conference on Informatics and Computing (ICIC) · 2022
The purpose of this study is to solve the problem of unbalanced data for prediction and classification of fraudulent E-Commerce transactions. Data from Digital Commerce 360 in 2015 showed that fraud occurred as much as 35% of total e-commerce transactions. Quoted from Bisnis.com, based on the 2017 Fraud Management Insight report, this percentage of fraud can reduce consumer confidence. One method for predicting fraud is machine learning. Fraud data does not have a balance between data that is not fraudulent, causing the classification to be biased. So it is necessary to balance the data using the SMOTE algorithm. The results of the data balancing will be classified as fraudulent transactions using the Support vector machine, K-Nearst Neighbor, Naïve Bayes and C45 algorithms.