Develop a testbed for fraud detection in e-banking systems, requiring (1) realistic generation of accounts and transactions (fraudulent patterns as well as normal ones), (2) fraud detection basis in form of a scoring model, (3) a simulation system for the evaluation of particular configurations and weights. Results have to be measured in terms of false positives/negatives + financial impact + duration of recognition. Requires excellent knowledge in data science. Usage of AI tools is not allowed for this topic, except for data generation.
