Markup Optimization With Neural Networks for Nonlinear Demand

Solution Overview

Problem

Existing third-party delivery service providers struggle with optimizing pricing strategies that consider non-linear price-demand relationships across a variety of items, leading to suboptimal revenue generation and inefficiencies in managing delivery services.

Innovation Solution

A neural network-based system is employed to simulate and identify optimal markup policies by analyzing non-linear price-demand relationships for multiple items, using a multi-item price-volume model and evolutionary optimization algorithms to determine pricing strategies that maximize revenue and sales volume.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If linear regression models or MILP/QP solvers are used for pricing optimization, then computational simplicity is maintained, but accuracy in capturing non-linear price-demand relationships deteriorates

Engineering Contradiction:

Improvecomputational simplicityVSAvoidaccuracy in capturing price-demand relationships

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent replaces traditional mechanical/mathematical optimization systems (MILP, QP solvers) with a neural network-based system. The neural network learns non-linear price-demand relationships from historical data, substituting the need for complex mathematical programming while achieving superior accuracy in capturing non-linear patterns in consumer behavior and demand elasticity.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent transforms the optimization approach by changing from fixed linear parameters (in MILP/QP) to adaptive learned parameters in a neural network. The system dynamically adjusts pricing strategies by learning from historical data, allowing parameters to evolve and capture non-linear relationships that traditional methods miss.

Inventive Principle:

Principle #35Parameter changes

2Ease of manufacture

If traditional optimization methods are used, then implementation ease is maintained, but ability to model complex non-linear relationships deteriorates

Engineering Contradiction:

Improveease of implementationVSAvoidability to model non-linear relationships

Core Design Contradiction:

Ease of manufactureVSAdaptability or versatility

Solution Approach 1:

The patent replaces traditional mechanical/mathematical optimization systems (MILP, QP solvers) with a neural network-based system. The neural network learns non-linear price-demand relationships from historical data, substituting the need for complex mathematical programming while achieving superior accuracy in capturing non-linear patterns in consumer behavior and demand elasticity.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The neural network system performs self-learning and self-optimization by automatically training on historical pricing and demand data. Once trained, the system autonomously generates optimal pricing strategies without requiring manual intervention or complex setup, making it both versatile and relatively easy to implement.

Inventive Principle:

Principle #25Self-service

3Ease of operation

If linear demand assumptions are made, then forecasting simplicity is maintained, but revenue optimization potential deteriorates

Engineering Contradiction:

Improveforecasting simplicityVSAvoidrevenue optimization potential

Core Design Contradiction:

Ease of operationVSProductivity

Solution Approach 1:

The patent transforms the optimization approach by changing from fixed linear parameters (in MILP/QP) to adaptive learned parameters in a neural network. The system dynamically adjusts pricing strategies by learning from historical data, allowing parameters to evolve and capture non-linear relationships that traditional methods miss.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent replaces traditional mechanical/mathematical optimization systems (MILP, QP solvers) with a neural network-based system. The neural network learns non-linear price-demand relationships from historical data, substituting the need for complex mathematical programming while achieving superior accuracy in capturing non-linear patterns in consumer behavior and demand elasticity.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS12367509B2Markup optimization

Publication Date: 2025.07.22 SHIPT INC

AI summary

In general, this disclosure is directed to determining an optimal markup policy for a third party delivery service. One aspect is a method for dynamically optimizing prices by a third party delivery service, the method comprising retrieving item information for a plurality of items including pricing data, the plurality of items being sold by a retailer and available for delivery by the third party delivery service, simulating a plurality of markup policies with a neural network trained to identify non-linear price-demand relationships between the plurality of items, determining, as part of simulating the plurality of markup policies, an optimal markup policy for the third party delivery service, and providing a user interface to a customer computing device, the user interface displaying at least some of the plurality of items for delivery by the third party delivery service with prices reflecting the optimal markup policy.