PwC CZ Data science & AI and Financial risk modelling teams provide full range of services to support the back-end of your credit decision process. This discussion material contains the following four different kind of products and services.
We develop a scoring model, heart of your decision process, tailored to your needs.
Deliverable: custom scoring model tailored to your needs.
We enhance your scoring with innovative data sources, such as geographic features or Telco data.
Telco scoring pilot,
PwC GeoDataMart adaptation & scoring pilot.
Together with our partnered firm Taran we deliver end-to-end back-end implementation of your loan approval process.
Deliverable: deployed software application orchestrating the back-end of your credit underwriting and monitoring end-to-end.
We ensure your credit decision process is compliant with relevant regulation regarding credit underwriting and consumer protection.
Deliverable: regulatory advisory services supporting components 1 – 3.
Many of the benefits of our solution revolve around its modularity, flexibility and tooling enabling rapid redeployment and innovation without being too reliant on the IT department. This is especially beneficial in new segments such as online lending requiring frequent strategy updates.
Highly modular and flexible.
Set-up a custom strategy for your approval process from available modules.
Cost-effective – smart caching of external and costly data source in the data caller.
New data are only requested when needed.
Multiple scoring strategies for benchmarking and/or segmentation of the loan approval process.
Anti fraud check based on info from antifraud DB such as device ID, IP address.
Every input and output is saved and every historical decision is replicable and deterministic.
Interactive reporting module over SQL database for monitoring your online population and basis for upgrades.
Real-time and batch scoring in one unified engine – real time for BAU application process and batch for campaigns on existing population.
In-built cross-sell and up-sell capabilities based on defined strategies.
Supports risk based pricing and/or propensity based smart pricing.
Open source tech mitigating vendor lock-in. Easy deploys virtually independent on IT department.
A license type enabling you to modify the code and develop on top of it for your purposes.
Analytical utilities - suite of utilities for data scientist working with TaranDM.
Extract predictors for analytical work. Predictor form and binning embedded in strategy.
The architecture of the demo application is based on microservices deployed on Azure Kubernetes Service (AKS). Bellow you can see the diagram of how the individual components of this solution are connected.