The Challenge.

  • Detecting fraudulent behaviour is one of the most important problems in the banking industry.
  • Fraud detection models must be both robust and extremely fast to ensure quick reaction to a fraudulent transaction.
  • The fast-moving landscape of financial crime means algorithm must be able to quickly adapt and learn from new trends.

The Solution.

  • Designed and built a Machine Learning model to identify fraudulent credit card transactions.
  • Integrated the model into enterprise systems to create alerts in real time when detecting fraudulent behaviours.
  • Built a continuous feedback loop to continue training the Machine Learning model so always learns from the latest trends.

The Results.

  • Reduced costs by decreasing the number of fraudulent transactions allowed to go through.
  • Improved customer experience by more quickly identifying and blocking fraudulent behaviour, as well as reducing the number of cases where a transaction was falsely labelled as fraudulent.
  • Reduced call center costs by decreasing the volume of calls for claims related to fraud.
style="vector-effect: non-scaling-stroke;">