[M] Detecting and Explaining Potential Financial Fraud Cases in Invoice Data using Machine Learning

MASTER Assignment

Detecting and explaining potential financial fraud cases in invoice data using machine learning

Type : Master M-BIT

Period: Feb, 2020 - Jan, 2021

Student : Hamelers, L.H. (Lieke, Student M-BIT)

Date Final project: January 19, 2021

Thesis

Supervisors:

dr. M. Poel (Mannes) Researcher (Guest)

+31534893920

 m.poel@utwente.nl

Building: Zilverling 4102

Personal page

prof.dr. J. van Hillegersberg (Jos) Full Professor

+31534893513

 j.vanhillegersberg@utwente.nl

Building: Ravelijn 3414

Personal page

Abstract:

This research looks into the different possibilities of using unsupervised outlier detection algorithms to detect potential fraud cases in invoice data of the public sector. Next, it researches possible explanation mechanisms to explain the process and outcomes of the algorithm. This research designs and validates an explanation facility for financial auditors and identifies opportunities for using this facility.