Artificial Intelligence and Machine Learning in Retail Banking

Enhancing Fraud Detection and Customer Segmentation

Resumen

Retail banks face two connected pressures. Fraud losses keep rising as payments move online and criminals adapt, and customers expect offers and service that feel personal in a market where products are easy to compare. Artificial intelligence and machine learning promise to address both, yet many deployments stall because models are opaque, because staff do not trust or adopt them, or because evaluation ignores the extreme imbalance of fraud data. This paper presents an integrated study that covers the technical and the human side. On the technical side, a pipeline compares four classifiers, a decision tree, logistic regression, gradient-boosted trees, and a support vector machine, selects the best through held-out evaluation, improves it with cross-validated tuning, and explains its decisions with local and global attribution methods. A clustering analysis segments 25,000 customers by behaviour......