Tree-Based Model Risk Determination via Node Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining action codes in tree-based models are challenging due to the tree structure, which makes it difficult to assess how much each feature affects the score, leading to approximations rather than actual factors affecting end scores.
Innovation Solution
A risk determination system that assigns each decision node or feature in a tree-based model to a correlation or action code, initializes a risk sum, calculates the difference in risk between child and parent nodes, updates the risk sum, and determines the correlation or action code associated with the highest risk sum, thereby identifying the actual factors affecting the score.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing methods for determining action codes in tree-based models are used, then the process can be completed, but the results are approximations rather than actual factors affecting end scores
Solution Approach 1:
The patent segments the tree-based model into individual decision nodes and traces multiple paths from root to leaf nodes. By breaking down the complex tree structure into manageable segments (individual paths and nodes), the system can accurately track which features contribute to each decision, transforming an intractable complexity problem into a series of manageable path analyses that yield precise feature attribution.
2Measurement precision
If the system determines actual factors affecting scores in tree-based models, then measurement precision improves, but additional computations are required
Solution Approach 1:
The patent performs preliminary action by pre-computing and storing the feature-to-action-code mappings and risk sums during the model training or setup phase. When determining action codes for new predictions, the system simply retrieves pre-computed values and sums them, rather than re-analyzing the entire tree structure. This preliminary preparation eliminates redundant computations and achieves O(1) lookup time during inference.
3Adaptability or versatility
If the system generates accurate adverse action codes within the same computational framework, then integration efficiency improves, but the tree structure complexity makes feature impact assessment difficult
Solution Approach 1:
The patent creates a universal action code determination system that works across different tree-based models (random forests, gradient boosting, etc.) by implementing a generalizable path-tracing algorithm. The same core mechanism—tracing paths from root to leaf and summing feature contributions—applies uniformly regardless of the specific tree variant, enabling the system to adapt to multiple model types without requiring model-specific customization while accurately assessing feature impact.
Data Source
AI summary
The present disclosure describes systems and methods for determining correlation codes for tree-based decisioning models. In one embodiment, a method for determining correlation codes in a tree-based decision model includes: assigning each decision node in a tree-based decision model to a correlation code; initializing a risk sum for each correlation code; calculating, for all decision nodes in the tree-based decision model, a difference in risk between child nodes and respective parent nodes; updating the risk sum for each correlation code associated with the decision node used in the decision for the node; determining the feature with the highest risk sum; and determining the correlation code associated with the determined decision node.