Decision Engine Score Decomposition for Credit Risk Transparency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Credit scoring systems face limitations in explicitly factoring specific borrower attributes and compliance with regulations, hindering accurate risk assessment and decision-making in lending processes.
Innovation Solution
A supervised learning-based decision engine is developed to determine decision scores as linearly decomposed constituent components, using Random Forest or Gradient Boosted Tree ensembles, which facilitates objective and compliant risk assessments by identifying contributing attributes, and generates actionable reports.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional credit scoring models are used, then credit risk assessment can be performed, but the ability to explicitly factor in specific borrower attributes is limited
Solution Approach 1:
The patent segments the decision score into multiple additive components, each corresponding to a specific borrower attribute. This segmentation allows the model to explicitly factor in and separately evaluate individual attributes such as employment status, credit history, and income level, thereby preventing loss of attribute information while maintaining overall risk assessment accuracy.
Solution Approach 2:
The patent transforms the traditional monolithic credit score into a parameterized form where the decision score is expressed as a function of multiple attribute parameters. By changing the scoring structure from a single aggregated value to a decomposed parametric form, the system can explicitly track and evaluate the contribution of each borrower attribute to the overall risk assessment.
2Reliability
If machine learning algorithms are used to build credit models, then predictive power is improved, but generating compliant reports becomes complex
Solution Approach 1:
The patent segments the machine learning model's output into additive attribute-specific components. This segmentation enables the generation of compliant reports by clearly showing which attributes contributed to the decision, simplifying the reporting process while maintaining the predictive power of the underlying machine learning algorithm.
Solution Approach 2:
The patent introduces an intermediary layer between the machine learning algorithm and the reporting system. This intermediary decomposes the model's predictions into interpretable attribute contributions, serving as a mediator that translates complex algorithmic outputs into compliant, explainable reports without sacrificing predictive accuracy.
3Measurement precision
If ordinal ranking credit scores are used, then risk ordering is achieved, but explicit attribution of risk factors is lost
Solution Approach 1:
The patent segments the overall risk score into additive components attributed to specific borrower characteristics. This segmentation preserves the ordinal ranking capability for risk assessment while simultaneously providing explicit attribution of risk factors, allowing stakeholders to understand not just the risk level but also which attributes drove the assessment.
Solution Approach 2:
The patent changes the scoring parameter structure from a single ordinal rank to a decomposed parametric representation. This parameter change maintains the ability to order borrowers by risk level while adding the capability to explicitly identify and measure the contribution of individual risk factors to each borrower's overall risk profile.
Data Source
AI summary
System, apparatus, user equipment, and associated computer program and computing methods are provided for facilitating efficient decision-making with respect to a subject entity. In one aspect, a labeled training dataset containing N records respectively corresponding to N entities is provided for training a decision engine based on performing supervised learning. Responsive to receiving a plurality of attribute values for the subject entity requiring a decision relative to an estimate of a performance variable based on at least a portion of the attribute values, the trained decision engine is configured to determine a decision score as a function obtained as a set of linearly decomposed constituent components corresponding to the attribute values of the subject entity, thereby effectuating an objective determination of which attributes contribute to what portions of the decision score in a computationally efficient manner.