Toggle Main Menu Toggle Search

Open Access padlockePrints

Experiment Pre-Registration Predicting 3-D-Secure Fraud Detection Outcomes

Lookup NU author(s): Professor Thomas GrossORCiD, Aamir Ali

Downloads


Licence

This is the final published version of a report that has been published in its final definitive form by Newcastle University, 2018.

For re-use rights please refer to the publisher's terms and conditions.


Abstract

Background. 3-D Secure [1] is an XML-based identity federation protocol meant to authenticate credit card transactions on the Web. The system uses a fraud detection model with a number of features extracted from the Web Browser to decide whether the user needs to be challenged with a password authentication or not. Aim. We aim to investigate and quantify the impact of factors used by 3-D Secure in its fraud detection decision making process and to establish a Logistic Regression Classifier for likely outcomes. Method. In a repeated measures experiment, four different credit cards were used to run credit card transactions with two different Web sites. During that experiment the independent variables machine data, value, region, and website were systematically manipulated, taking two levels each. We measured whether the user was challenged with an authentication, whether the transaction status was declined or accepted, and whether the card was blocked as nominal dependent variables. We establish logistic regressions as predictive model how the independent variables impact the likelihood of the response variables changing. We also seek to establish a multiple logistic regression as overall model. Anticipated Results. We anticipate the logistic regressions to yield odds ratios as quantification for the impact of individual independent variables on the response variables. Anticipated Conclusions. We anticipate that the logistic regression models are insightful in explaining the (probabilistic) security and fraud-detection decisions taken at the back-end.


Publication metadata

Author(s): Gross T, Ali MA

Publication type: Report

Publication status: Published

Series Title: School of Computing Technical Report Series

Year: 2018

Pages: 9

Print publication date: 01/12/2018

Acceptance date: 01/01/1900

Report Number: 1537

Institution: Newcastle University

Place Published: Newcastle upon Tyne

URL: https://www.ncl.ac.uk/media/wwwnclacuk/schoolofcomputingscience/files/trs/1537.pdf


Share