Constrained Regression Modeling for Reliable Quality Coefficients
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing regression models in production systems struggle to maintain reliability and accuracy due to variations in data trends and the influence of noise, often leading to coefficients that deviate from expert perception, thereby reducing the model's effectiveness in quality management.
Innovation Solution
The information processing device constrains model coefficients within predefined ranges based on expert domain knowledge and historical data to ensure the coefficients align with expected values, using techniques like setting constraints on coefficient ranges and optimizing models to maintain accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If regression models are constructed using machine learning to analyze process data, then the ability to identify primary factors of quality variation is improved, but the reliability and accuracy of the model coefficients deteriorate when data trends vary or noise is present
Solution Approach 1:
The patent changes the parameter constraints by imposing range constraints on regression coefficients based on expert knowledge. This transforms the unconstrained coefficient estimation problem into a constrained optimization problem where coefficients are restricted to fall within predefined ranges, thereby improving reliability while maintaining accuracy
Solution Approach 2:
The patent incorporates feedback mechanisms by using expert domain knowledge to define constraint ranges for coefficients. This feedback loop ensures that the model coefficients remain within theoretically and practically meaningful boundaries, preventing divergence due to noise or data trend variations
2Measurement precision
If the model is optimized to fit the data, then the accuracy improves, but the coefficients may deviate from expert perception and historical data, reducing model effectiveness
Solution Approach 1:
The patent introduces constraint ranges as an intermediary between the data-driven optimization process and expert knowledge. These ranges act as a mediator that allows the model to fit the data while ensuring coefficients remain consistent with expert perception and historical data, preventing information loss
Solution Approach 2:
The patent performs preliminary action by pre-defining constraint ranges for coefficients based on expert knowledge and historical data before conducting the regression analysis. This preliminary constraint setting ensures that the optimization process remains anchored to domain knowledge throughout the modeling process
Data Source
AI summary
According to an embodiment, an information processing device includes one or more processors. The one or more processors are configured to: set, for each of a plurality of parameters corresponding to a plurality of explanatory variables included in a first model that inputs the plurality of explanatory variables and performs estimation, a constraint condition including a first range of a value of the corresponding parameter; and construct the first model by obtaining the plurality of parameters that satisfy the constraint condition.