Unified Regression Analysis System for Simultaneous Discrimination and Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current regression analysis methods in industrial fields, such as the medical field, face challenges in accurately estimating evaluation values due to inconsistent discrimination between patient and unimpaired groups, leading to incomplete evaluation and reduced accuracy, especially with limited data samples.
Innovation Solution
A unified single index is used to express evaluation values for both groups, allowing simultaneous discrimination and evaluation, enabling the estimation of evaluation values for all samples and improving accuracy by incorporating additional units like evaluation value conversion, priority adjustment, convergence determination, and important characteristic selection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a two-step discrimination and regression process is used, then the evaluation value can be estimated for samples in the estimable group, but the evaluation value cannot be estimated for samples mistakenly discriminated as inestimable, reducing operational completeness
Solution Approach 1:
The patent combines the discrimination process and regression process into a single unified process. The regression model incorporates a discrimination function that simultaneously determines group membership and estimates evaluation values, eliminating the need for sequential processing and ensuring all samples receive evaluation estimates.
Solution Approach 2:
The regression model is designed to serve multiple functions: it performs both discrimination (classifying samples into estimable/inestimable groups) and regression (estimating evaluation values) within a single framework, making the system universally applicable to all samples regardless of their initial classification.
2Measurement precision
If a two-step discrimination and regression process is used, then the evaluation value can be estimated for samples in the estimable group, but the overall accuracy is reduced due to sequential processing and data loss
Solution Approach 1:
The patent merges the discrimination function and regression function into a single integrated model. The loss function combines both discrimination accuracy and regression accuracy objectives, allowing the system to optimize both simultaneously rather than sequentially, thereby improving overall evaluation accuracy while reducing process complexity.
Solution Approach 2:
The discrimination boundaries and regression parameters are determined simultaneously in advance through unified training, rather than performing discrimination first and then regression on the selected samples. This preliminary unified action ensures consistent and accurate results for all samples.
3Quantity of substance
If separate discrimination and regression standards are obtained, then the process can handle distinct classification and estimation tasks, but the number of required data samples increases, reducing data efficiency
Solution Approach 1:
The unified regression model is designed to perform both discrimination and regression tasks using the same set of data samples. By formulating the discrimination boundaries as part of the regression optimization problem, the model achieves dual functionality without requiring separate training datasets, thereby improving data efficiency while maintaining the flexibility to handle both classification and estimation tasks.
Data Source
AI summary
The present invention solves a problem that there may be a case that an estimated value of regression cannot be calculated depending on a discrimination result when a regression method is applied after a discrimination method, and has a purpose to obtain an estimating equation with high accuracy even when the number of sample groups to which the regression method is applied is small. An estimating equation that satisfies the regression and discrimination at the same time can be obtained by combining a discrimination evaluation function that evaluates discrimination accuracy and a regression evaluation function that evaluates regression accuracy, calculating a combination evaluation function, and optimizing the combination evaluation function.