Multi Sourced Financial Record Analytics and Fraud Detection Big Data

  • Bader Mohammed

Student thesis: Master's Thesis

Abstract

Big data analytics for fraud detection by analyzing multiple financial records of individuals, collected from various sources such as financial institutions (banks and insurance) and government departments within UAE. And develop an algorithm to solve the detected issue to prevent the fraud. 1. Data mining to discover people suspected for fraud a. Build a warehouse out of multiple databases, based on records received from Bank (16 banks), government entities such as Emirates Identity Authority, Housing Authority, Municipal Affairs, Department of Naturalization & Residency and 16 more. 2. Identify people who have launched 2 or more successful applications and granted loan a. Multiple criterion to come up with list of applications launched by applicants b. Application by date and comparison c. Analysis based on financial fraud statistics 3. Detect cash transaction under country regulation set thresholds a. Process to acquire records below regulatory thresholds to perform detect b. Record number of foreign exchange transfers below AED 4000 and look for individuals requested loans 4. Identify forged checques and loan applications a. Perform background checks with-in the records colonized with the data warehouse. b. Comparison of income and spending records, to justify the loan repayment 5. Make processes & procedures to deter fraud a. Credit grant process to be introduced to banks to take better decisions b. Process to decrease workload of forged application c. Blacklist creation for falsified candidates to be used and link other processes d. Blacklist of defaulters mapped with nationality and range of income.
Date of AwardNov 2016
Original languageAmerican English
SupervisorErnesto Damiani (Supervisor)

Keywords

  • Big Data; Fraud; Detection; Data Mining; Financial Institutions.

Cite this

'