Loan Fraud Detection Using Machine Learning as a Data Mining Approach

Loan Fraud Detection Using Machine Learning as a Data Mining Approach

Nabila Hamdoun
Copyright: © 2022 |Pages: 10
DOI: 10.4018/IJDA.309096
OnDemand:
(Individual Articles)
Available
$37.50
Current Special Offers
No Current Special Offers
TOTAL SAVINGS: $37.50

Abstract

Detecting loan fraud is a subject frequently raised in scientific research, and to which financial institutions attach great importance, given the losses they avoid by applying an efficient fraud detection. The purpose of this paper is to compare a machine learning algorithm with statistical models in fraud detection and describe a machine learning approach to loan fraud detection using data mining techniques. The paper proposes an approach using data mining techniques for machine learning in order to improve performance. In the training stage, the authors used two datasets. The first one is a raw data, and the second is a pre-processed data with feature engineering and selection methods. The results show a significant improvement in performance of all classifiers once tested on pre-processed data. Finally, they conclude that applying machine learning algorithms directly to raw data gives bad results for statistical model comparing to machine learning algorithms, and the proposed approach using feature engineering and selection techniques contributes to improve performance.
Article Preview
Top

1.1 Machine Learning In Fraud Detection

In a previous study (Hamdoun & Rguibi, 2019), we detailed the application of machine learning in Credit risk scoring by comparing between the statistical model Logistic Regression and the Random forest algorithm. The relevance and effectiveness of machine learning methods have also enabled financial institutions and Insurance, to optimize many other risks and detection of fraudulent behaviors, such as fraudulent Insurance claims, anti-money laundering and counter-terrorist financing (AML-CFT), Electronic payment card fraud, or Credit fraud, etc.

Complete Article List

Search this Journal:
Reset
Volume 5: 1 Issue (2024)
Volume 4: 1 Issue (2023)
Volume 3: 2 Issues (2022): 1 Released, 1 Forthcoming
Volume 2: 2 Issues (2021)
Volume 1: 2 Issues (2020)
View Complete Journal Contents Listing