University of Twente Student Theses

Login

Improving payment authorization rates : by intelligently routing transactions

Ciobotea, C. (2016) Improving payment authorization rates : by intelligently routing transactions.

[img]
Preview
PDF
4MB
Abstract:The payment industry has closely followed and benefited from the new technologies and e-commerce trends. Easy, fast and flexible payment methods have become a necessity. Nevertheless, payments are not always successful and as much as 20% of payments may fail. The research goal is to improve the authorization rate of genuine payments by designing an optimization algorithm for routing transactions in the payment flow, algorithm that supports unknown and unfixed authorization rates, such that payments are more efficiently processed. In other words, the algorithm should be able to choose the optimal acquirer to process payments with from a set of two or more acquirers, for each new transaction that comes in. We analyze several possible solutions, such as: Play the Winner Rule (randomized experimental design) and 4 multi-armed bandit strategies: e-greedy and pursuit strategy, gradient bandit, Exp31 and Thompson sampling (reinforcement learning). These algorithms are applied on real payment data. Simulations have shown that these algorithms have the potential of bringing an uplift of 1.54% to 2.84% in authorized transactions. In this thesis, we also evaluate the performance of the algorithms for different input parameters.
Item Type:Essay (Master)
Clients:
Adyen B.V., Amsterdam, Netherlands
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Computer Science MSc (60300)
Link to this item:http://purl.utwente.nl/essays/70977
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page