Partial Dependence Plots for Dynamic Service Level Optimization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional IT systems face challenges in meeting sophisticated service level metrics (SLMs) such as user satisfaction or dissatisfaction, as they often rely on overprovisioning, which leads to resource waste and inefficiency, and are unable to effectively understand and predict the driving factors behind user satisfaction, especially in diverse user groups.

Innovation Solution

The system uses machine learning and partial dependence plots (PDPs) to determine high-impact features on composite SLMs like user satisfaction, predicts their marginal contribution, and intelligently regulates IT system resources to maintain target SLM levels through alert generation and resource allocation, ensuring consistent user satisfaction across different user segments.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If overprovisioning is used to meet service level requirements, then service level delivery is improved, but resource waste increases

Engineering Contradiction:

Improveservice level deliveryVSAvoidresource waste

Core Design Contradiction:

ReliabilityVSLoss of energy

Solution Approach 1:

The system dynamically adjusts resource allocation based on real-time analysis of partial dependence plots and machine learning models, transitioning from static overprovisioning to adaptive resource management that responds to actual service level needs

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes the parameters of resource allocation by using machine learning-derived thresholds and partial dependence analysis to optimize resource distribution, moving away from fixed overprovisioned levels to data-driven dynamic parameters

Inventive Principle:

Principle #35Parameter changes

2Reliability

If overprovisioning is used to meet service level requirements, then service level delivery is improved, but system complexity increases

Engineering Contradiction:

Improveservice level deliveryVSAvoidsystem complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system introduces machine learning models and partial dependence plots as intermediaries between raw service level data and resource allocation decisions, simplifying the complex relationship between multiple service level metrics and resource requirements

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system replaces manual or rule-based resource allocation mechanisms with automated machine learning-based threshold determination and dynamic adjustment systems

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Ease of operation

If conventional IT systems use blanket thresholds for service level metrics, then implementation is simplified, but user satisfaction consistency deteriorates

Engineering Contradiction:

Improveimplementation simplicityVSAvoiduser satisfaction consistency

Core Design Contradiction:

Ease of operationVSStability of the object's composition

Solution Approach 1:

The system applies local quality by determining service level thresholds specific to different user segments and service contexts through partial dependence analysis, rather than using uniform blanket thresholds across all scenarios

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The system segments the user base and service metrics into distinct groups analyzed through partial dependence plots, allowing tailored threshold determination for each segment while maintaining overall system coherence

Inventive Principle:

Principle #1Segmentation

4Productivity

If machine learning and partial dependence plots are used to determine service level thresholds, then resource allocation efficiency is improved, but system complexity increases

Engineering Contradiction:

Improveresource allocation efficiencyVSAvoidsystem complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The system enables self-service by allowing the machine learning models to automatically determine thresholds and generate recommendations without requiring deep expert intervention, with the partial dependence plots providing intuitive visual explanations

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS20220187969A1Optimizing Service Delivery through Partial Dependency Plots

Publication Date: 2022.06.16 CERNER INNOVATION INC

AI summary

This disclosure includes technologies for service level delivery, including for achieving various threshold satisfaction levels in delivering services. The disclosed system uses machine learning models to predict the respective importance of various variables associated with a service. Further, the disclosed system determines respective marginal contributions and respective thresholds associated with variables with high-impact for service level delivery. Subsequently, the disclosed system performs various tasks based on those thresholds to achieve various threshold satisfaction levels in delivering the service.