FRAUD DETECTION ON MOBILE MONEY TRANSFER USING DECISION TREE AND SUPPORT VECTOR MACHINE

• Main Author: Viltoriano, Robin
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/64940
• Institution: Institut Teknologi Bandung

Mobile money transfer is a digital financial transaction activity that is performed on the user's smartphone. Mobile money transfers have several advantages over traditional transactions because they can be completed instantly anywhere and at any time. This digitalization has different security standards than traditional transactions. Weak cyber security creates opportunities for money theft by allowing fake transactions to take place at any time and anywhre. As a result, the goal of this research is to create a model that can determine whether a transaction is fraudulent. Historical data is required for the learning process when creating a supervised learning model. Since financial transaction datasets are scarce, we'll use a synthetic dataset. PaySim data was used, which is a simulation of mobile money transfer data based on actual transaction data. Decision Tree and Support Vector Machine models, which are enhanced with the SMOTE-Tomek Link method, will be used to detect fake transactions. Based on the AUC and F1-Score metrics, we will compare the effects of adding the SMOTE-Tomek Link method to the performance of the two models. When compared to other models, the Decision Tree model with the addition of the SMOTE-Tomek Link method appears to be the best model for this dataset, according to the results of the experiment.