Scoring and explainability
A model is assembled from rules and features, computes a score and returns not just the result but the list of factors that shaped it, with the contribution of each.
What is inside
- 01
Engines: rules, logistic regression, ONNX model, external service and a hybrid scheme.
- 02
Features come from the form, the product, in-house history and credit bureau reports.
- 03
Rule actions: points, decline, approve, manual review, cap on amount, term and rate.
- 04
Every decision stores the factors and how much each contributed to the final score.
Nearby in this group
Leads and sources
Leads with campaign tags before the application: assignment, contacts, qualification and conversion into a customer.
Applications and pipeline
One application pipeline across every channel with configurable check steps.
Credit bureaus
Report requests and reverse publication of contract events to the bureau of each country.