Distributed Rate Adjudication Using Double Machine Learning

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Organizations often impose uniform pricing or rate treatments across customers, leading to underutilization of computational resources due to lack of personalization based on customer sensitivity, and existing predictive processes are inaccurate and computationally infeasible for characterizing rate sensitivity across a continuum.

Innovation Solution

Implementing a double machine-learning process that includes de-noising, de-biasing, and linear regression to predict customer-specific treatment elasticity, coupled with a trained classifier to label customers as low- or high-treatment-elasticity, facilitating real-time adjudication of exception requests.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If uniform pricing or rate treatments are imposed across customers, then organizational simplicity and ease of operation are improved, but resource utilization deteriorates due to lack of personalization

Engineering Contradiction:

Improveease of imposing uniform pricingVSAvoidresource utilization

Core Design Contradiction:

Ease of operationVSProductivity

Solution Approach 1:

The system transitions from uniform pricing to personalized pricing by applying different rate treatments to different customer segments based on their treatment elasticity. The double machine learning process and classifier enable local optimization of pricing strategies for each customer or customer group, matching pricing sensitivity to individual customer characteristics while maintaining organizational manageability through automated classification.

Inventive Principle:

Principle #3Local quality

2Adaptability or versatility

If existing predictive processes are used to characterize rate sensitivity, then some level of customer segmentation is achieved, but accuracy and computational feasibility deteriorate

Engineering Contradiction:

Improvecustomer segmentation capabilityVSAvoidaccuracy of rate sensitivity characterization

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The system replaces traditional mechanical or rule-based predictive processes with a double machine learning approach. This substitution enables more accurate characterization of treatment elasticity by using trained classifiers and regression models that can capture complex, non-linear relationships between customer attributes and pricing sensitivity, significantly improving measurement precision while remaining computationally feasible.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Adaptability or versatility

If traditional predictive processes are used for rate sensitivity analysis, then some predictive capability is achieved, but computational efficiency deteriorates due to infeasibility at scale

Engineering Contradiction:

Improvepredictive capabilityVSAvoidcomputational efficiency

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The system performs preliminary classification of customers into treatment elasticity segments using trained classifiers before applying detailed predictive analysis. This preliminary action groups customers with similar characteristics and pricing sensitivities, enabling more efficient computational processing at scale by avoiding the need to perform complex analyses on every individual customer separately.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The double machine learning process segments the customer base into distinct groups based on treatment elasticity characteristics. This segmentation enables the organization to apply different pricing strategies and rate treatments to different segments, improving both predictive accuracy and computational efficiency by handling segmented data rather than attempting to model all customers uniformly.

Inventive Principle:

Principle #1Segmentation

4Stability of the object's composition

If standardized parameter values are used for pricing, then consistency and ease of operation are improved, but personalization and customer relationship optimization deteriorate

Engineering Contradiction:

Improveconsistency of pricing parametersVSAvoidpersonalization capability

Core Design Contradiction:

Stability of the object's compositionVSAdaptability or versatility

Solution Approach 1:

The system introduces dynamic pricing capabilities while maintaining operational consistency through automation. The double machine learning process and trained classifiers enable pricing parameters to adapt dynamically to individual customer characteristics and treatment elasticity, allowing the organization to move from static standardized pricing to dynamic personalized pricing without sacrificing consistency, as the automated system applies rules uniformly across all customers based on their segment classification.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS20250307660A1Causal inferencing in distributed computing environments using trained double machine learning and trained classifiers

Publication Date: 2025.10.02 THE TORONTO DOMINION BANK

AI summary

The disclosed embodiments include computer-implemented apparatuses and processes that perform causal inferencing in distributed computing environments using trained double machine learning and trained classifiers. For example, an apparatus may receive, from a device, a request that includes identifier associated with the device and exception data that includes a requested modification to a value of a parameter of a data exchange. The apparatus may also obtain labelling data based on an application of a trained classifier to a first input dataset that includes a value of an elasticity parameter associated with the request, may generate elements of decision data associated with the requested modification based on the labelling data and on the exception data, and may transmit, to the device, a response to the request that includes the elements of decision data.