Credit Scoring

Operation keyword for Features: features
Operation keyword for Credit Scores: credit-scores

Credit Scoring is a set of customisable analytical solutions for Enterprise-level clients of Nordigen that require the most advanced capabilities. Credit Scoring set includes two solutions: Features and Credit Scores.

Credit Scoring includes access to a library of up to 1 million predictive machine learning features (modelling parameters), custom model training services and bespoke infrastructure deployments.

Request Flow

To use this product via API, see the Endpoints section "report" and go through the following steps:

Upload the account information;
Apply operation features or credit-scores;
Get results.

Features

Features are representations from transaction data that are used as an input for scoring models.

Response Example


{
    "data": {
        "attributes": {
            "features": {
                "month_avg_count_cat_23_3m": {
                    "b5db8f2f-bccc-4be3-a296-6a430cc347c8": 0.00333
                },
                "month_avg_count_cat_84_3m": {
                    "b5db8f2f-bccc-4be3-a296-6a430cc347c8": 0.01
                },
                "month_avg_count_cat_85_3m": {
                    "b5db8f2f-bccc-4be3-a296-6a430cc347c8": 0.01
                }
            },
            "status": "completed"
        },
        "type": "report processing status"
    }
}

Key Descriptions

Key	Description	Description
<function or feature logic name>_cat_<category_id>_<months>m (e.g.median_cat_79_12m)	Float	Numerical representations (features) of statistically transformed raw transaction data on the aggregated account level. Each feature is associated with specific transformations on specific categories or category groups
<statement_id> (e.g. abcdefg12345)	String	Unique identifier for each statement
status	String	Processing record state
type	String	Response type

Credit Scores

Credit scoring means calculating the probability of a default, based on behaviours identified in transaction data.

Response Example


{
    "data": {
        "attributes": {
            "credit-scores": {
                "model_12m_1v.model": 0.544434130191803,
                "model_6m_3v.model": 0.32850202918052673
            },
            "status": "completed"
        },
        "type": "report processing status"
    }
}

Key Descriptions

Key	Type	Description
model-name (e.g. model_test_v12.0.model)	Float	A probability that the statement is going to default
status	String	Processing record state
type	String	Redundant field, it will be removed in the next iteration