Akhmad, Arsandi and Lukas, Lukas and Mahawan, Bagus (2019) Combining Loan Fraud Detection and Deposit Risk Profiles to Improve the Effectiveness of Deposit Reconciliation and Verification Process in IDIC. Masters thesis, Swiss German University.

Preview	Text Arsandi Akhmad 21751008 TOC.pdf Download (1MB) | Preview
	Text Arsandi Akhmad 21751008 1.pdf Restricted to Registered users only Download (761kB)
	Text Arsandi Akhmad 21751008 2.pdf Restricted to Registered users only Download (324kB)
	Text Arsandi Akhmad 21751008 3.pdf Restricted to Registered users only Download (1MB)
	Text Arsandi Akhmad 21751008 4.pdf Restricted to Registered users only Download (351kB)
	Text Arsandi Akhmad 21751008 5.pdf Restricted to Registered users only Download (387kB)
Preview	Text Arsandi Akhmad 21751008 Ref.pdf Download (297kB) | Preview

Abstract

IDIC as banking regulator in the mandate to handle the failed bank has two main tasks, namely completing all bank obligations or known as the liquidation of the bank and guaranteeing deposits that exist in the bank. In carrying out its duties as deposit guarantor, IDIC conducts a series of processes called reconciliation and verification. This study discusses how to increase the effectiveness of the reconciliation process and the verification of deposits doing by IDIC. Usually, this process is complicated because most banks that fail and liquidated because fraud and the most significant portion of fraud in the failed bank is loan fraud. The existence of fraud makes the validity of the balance recorded on the core banking application doubtful. It was often found in the flow of customer deposits that funds originate from loan fraud and to ensure that it takes a long time while the period of the reconciliation and verification process is limited to ninety working days. For this reason, IDIC need a method that can increase effectiveness in this process, especially related to the detection of fraud. So in this study, a method was developed by using depositors' risk profiles combined with the detection of loan fraud that aims to increase the effectiveness of reconciliation and verification process in order to insured deposit. In developing this method, data is needed following the process of reconciliation and verification of deposits as noted as in Standard Operation Procedure such as data related to savings, deposits, loans and detailed data related to transactions. These data are processed and make the dataset in this study process and will use in loan fraud detection and depositors’ risk profile. This application uses the CRISP-DM framework with the Random Forest algorithm as the method. In this study also uses the SMOTE algorithm as handling data imbalance. The results of the analysis show that the features extracted or selected from failed bank data can be used in predicting loan fraud and insured deposit predictions with high accuracy and precision.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Data Analytics ; Machine Learning ; Predictions ; Banks ; Fraud
Subjects:	T Technology > T Technology (General) > T58.6 Management information systems
Divisions:	Faculty of Engineering and Information Technology > Department of Information Technology
Depositing User:	Adityatama Ratangga
Date Deposited:	19 May 2020 04:13
Last Modified:	19 May 2020 04:13
URI:	http://repository.sgu.ac.id/id/eprint/493

Actions (login required)

View Item