FRAUD DETECTION USING PARTICLE SWARM OPTIMIZATION-BASED SUPPORT VECTOR MACHINE

• Main Author: Denaya Rahadika Diana, Kadek
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/38979
• Institution: Institut Teknologi Bandung

Fraud is a criminal deception committed by Bank or E-commerce user intended to result in personal gain. Fraud detection is one of an important thing that the Bank and E-commerce must be concerned about. Fraud must be able to be detected by Bank or E-commerce to minimize losses caused by this action. Support Vector Machine or SVM is a classification method in machine learning which can be used to classify a transaction as a fraud or not fraud. To use SVM, it requires estimation of several hyperparameters values to get the best result. Particle Swarm Optimization or PSO is a swarm-based optimization method which can be used to search the best value for a specific objective function. In this research, the author used PSO to find the best hyperparameter values of SVM that can give the best model performance and able to classify the transaction data as fraud or not.