Making Sense of Bank Networks Using High-dimensional Classification Techniques

We build an integer-programming optimization model to classify the entities of a directed network in a high-dimensional feature space. The network is associated with cross-border payments flowing through the global SWIFT platform. Our objective is to detect, in the network of an international commercial bank, the low-profit high-risk entities, so they can be removed without losing a lot of business.

Ivan Sergienko is Director, Data Science & Analytics at Scotiabank and an Affiliated Researcher at Scotiabank Centre for Customer Analytics at Smith School of Business. He has ten years of banking experience in various roles in modelling, risk management, and customer analytics. Prior to his banking career, and after graduating with Ph.D. in Physics, he spent five years doing theoretical and computational research at Oak Ridge National Lab and Memorial University of Newfoundland. Ivan also holds a Chartered Financial Analyst® designation.