Optimizing the financial gain for credit card fraud detection systems using machine learning techniques Rik Vegter

Vegter, Rik (2022) Optimizing the financial gain for credit card fraud detection systems using machine learning techniques Rik Vegter. Master's Thesis / Essay, Artificial Intelligence.

Preview	Text Master_Thesis__Credit_Card_Fraud_Detection.pdf Download (1MB) | Preview
	Text toestemming.pdf Restricted to Registered users only Download (156kB)

Abstract

Credit card fraud detection is a worldwide problem that comes with significant financial losses for credit card service companies. Due to the high number of transactions that occur daily, there is a need for the automatization of the classification of these transactions using machine learning techniques. One of the main problems within credit card fraud detection is the imbalance between fraudulent and genuine transactions. Machine learning algorithms tend to be biased towards the majority class. Therefore data balancing techniques are required to overcome the problem of class imbalance. This thesis provides a comparison of state-of-the-art methods in terms of a new proposed, more realistic financial metric. The financial metric is defined based on expert advice from icscards and the litera- ture of credit card fraud detection. While the financial metrics in literature only take into account a percentual gain and loss of transactions, this metric also takes external costs that credit card service companies make into account. Also, a class imbalance machine learning method used to optimize the decision threshold from another domain (called Ghost) where class imbalance occurs is trained on credit card fraud data. Ghost finds the optimal decision threshold combined with a Random Forest classifier. The results were tested on two commonly used credit card fraud detection data sets and showed that boosting methods perform state-of-the-art results both in terms of the number of cor-

Item Type:	Thesis (Master's Thesis / Essay)
Supervisor name:	Dhali, M.A. and Schomaker, L.R.B.
Degree programme:	Artificial Intelligence
Thesis type:	Master's Thesis / Essay
Language:	English
Date Deposited:	14 Nov 2022 10:09
Last Modified:	14 Nov 2022 10:09
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/28941

Actions (login required)

View Item