The bank needed to build a risk assessment model for customers by identifying viable risk factors, including customer profile, behavioral/transactional attributes, and payment services/delivery channels. The objective was to compute a customer risk rating (CRR) score to help mitigate AML risk.

Developed a comprehensive CRR model using Bayesian network and Fuzzy Set theory.

1. Data Review and Validation: Standardized and mapped customer and transaction data, identified key attributes for both business and individual customers, and performed data validation and quality checks.

2. Calculation of CRR Factor Scores: Computed individual factor scores at the account level and aggregated them at the customer level. Assigned weights to each factor and calculated the final CRR score by combining weights with scores. Defaulted MSB, PEP, and SAR customers to a high-risk score.

3. Model Development: Used Bayesian network (directed tree graph) and Fuzzy Set theory to classify customers into low, medium, and high-risk buckets.

4. Automation and Deployment: Implemented the model to classify customers in various risk buckets automatically.

Automated classification of customers into risk buckets.

The model allowed the bank to classify customers into different risk categories in an automated manner, enhancing the efficiency and effectiveness of their AML risk mitigation efforts.