Presented By
O’Reilly + Cloudera
Make Data Work
29 April–2 May 2019
London, UK
Please log in

Predicting real-time transaction fraud using supervised learning

Sami Niemi (Barclays)
11:1511:55 Wednesday, 1 May 2019
Data Science, Machine Learning & AI
Location: Capital Suite 17
Average rating: ****.
(4.62, 16 ratings)

Who is this presentation for?

  • Data scientists, machine learning enthusiasts, and financial and model risk experts



Prerequisite knowledge

  • Familiarity with machine learning and supervised techniques (useful but not required)

What you'll learn

  • Learn how real-time transaction fraud models work, the main challenges in transaction fraud modeling, which supervised machine learning techniques are most applicable, and what are the future research areas (e.g., unsupervised and sequence models)


Predicting transaction fraud of debit and credit card payments in real time is an important challenge, which state-of-the-art supervised machine learning models can help to solve. While supervised learning techniques, like logistic regression and neural networks, have been used for many years, recent developments in deep learning, gradient-boosted machines, and recurrent neural networks have opened up a wealth of options that can provide significant improvements over the existing models. These techniques are in general well suited for transaction fraud, but large data volumes (billions of transaction per year), very imbalanced target classes, ever-changing fraud MOs, and strict requirements for the prediction inference speed mean that some methods are better suited than others.

Sami Niemi offers an overview of the solutions Barclays has been developing and testing and details how well models perform in variety of situations like card present and card not present debit and credit card transactions. Sami demonstrates how to train supervised transaction fraud models that can be implemented and how these models improve both customer experience and help to reduce fraud losses. He then explores results of a machine learning model that is operating in production.

Photo of Sami Niemi

Sami Niemi


Sami Niemi is a vice president and managing data scientist at Barclays, where he leads a team of data scientists building fraud detection models and manages the UK fraud models. Sami has been working on Bayesian inference and machine learning for over 10 years and has published peer-reviewed papers in astrophysics and statistics. He has delivered machine learning models for telecommunications and financial services and built supervised learning models to predict customer and company defaults, first- and third-party fraud, and customer complaints, and used natural language processing for probabilistic parsing and matching. He has also used unsupervised learning in a risk-based anti-money-laundering application.