Otkrivanje prijevara u korištenju kreditnih kartica primjenom metoda strojnog učenja

master's thesis

• Title (english): Credit Card Fraud Detection Using Machine Learning Techniques
• Author: Antoni Dragovčić
• Mentor: Goran Delač (mentor)
• Committee member: Goran Delač (predsjednik povjerenstva)
• Committee member: Stjepan Groš (član povjerenstva)
• Committee member: Siniša Srbljić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2019-07-09, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Tema ovog rada je razvoj programskog rješenja za otkrivanje prijevara u kartičnom plaćanju, kao jednom od rastućih problema u financijskom sektoru. U tu se svrhu koriste popularni algoritmi strojnog učenja -- neuronska mreža tipa autoenkoder i logistička regresija. Implementirana su tri pristupa problemu. Prvi koristi autoenkoder, drugi logističku regresiju, dok treći pristup kombinira te dvije metode. Dodatno, navedeni algoritmi su potkrijepljeni teorijskom podlogom. Rješenja se vrednuju nad neuravnoteženim skupom podataka stvarnih transakcija prikupljenih u razdoblju dva dana 2013. godine na europskom platežnom području. Korištene metrike vrednovanja su preciznost, odziv i mjera f1. Konačno, dobiveni rezultati se komentiraju i uspoređuju.

Abstract (english)

The topic of this thesis is the development of a software solution for detecting frauds in card payments as one of the growing financial sector issues. For this purpose, popular machine learning algorithms -- autoencoder as a type of neural network and logistic regression are being used. Three approaches to the problem were implemented. The first one uses an autoencoder, the second one logistic regression and finally, the third approach combines the first two methods. In addition, these algorithms are supported by theoretical background. Solutions are valued over an unbalanced set of real transaction data collected in the two-day period of 2013 on the European Payments Area. The metrics used are precision, recall and f1-score. Finally, the measured results are compared and commented on.

• Keywords: otkrivanje prijevara u kartičnom plaćanju; strojno učenje; autoenkoder; logistička regresija
• Keywords (english): credit card fraud detection; machine learning; autoencoder; logistic regression
• Language: croatian
• URN:NBN: urn:nbn:hr:168:283661
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2020-01-07 17:42:51