Overview
Every loan application is assessed by one or more scoring providers before a credit decision is made. HES LoanBox supports both the internal engine and the GiniMachine engine out of the box:
| Provider | What it is | Where the logic lives |
|---|---|---|
| Internal | A rule-based decision table (DMN) authored by the lender’s credit officer. Returns an internal rank. | LoanBox — Settings, Scoring model |
| GiniMachine | A machine-learning model trained on the lender’s historical loan data. Returns a Probability of repayment, a resolution and a per-attribute explanation. | The connected GiniMachine instance |
The provider outputs are not used directly. They are passed into a second decision table (Scoring result), which reconciles them into a single verdict called the Result Probability of repayment: green zone, grey zone, or red zone. The credit officer sees that verdict when making a decision.
The providers are deliberately independent. The internal model encodes policy that the lender can read, audit, and change on demand. GiniMachine encodes patterns learned from the portfolio’s own repayment history.
Running both means a lender gets a machine-learning signal without giving up a human-readable basis for the decision, and disagreement between the two becomes a routing signal rather than a problem.
Scoring is configured in the Decision making block of the product group form, not globally. Two product groups on the same platform instance can use different providers and different GiniMachine models.