Feature Contribution Score Classification Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Non-expert users face challenges in interpreting feature contribution scores from machine learning models due to their numeric nature and variability across different models, making it difficult to understand the influence of input features on the target feature.
Innovation Solution
An intelligent supervised feature contribution category classification model that predicts accurate and consistent categorical labels for feature contribution scores, independent of the machine learning model used, allowing for flexible application across various feature contribution set sizes and enabling interpretable insights for non-expert users.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If feature contribution scores are presented in numeric form, then measurement precision is improved, but ease of operation deteriorates for non-expert users
Solution Approach 1:
The patent applies color-coded visual indicators to represent different levels of feature contribution scores. High contribution scores are displayed with one color intensity, medium scores with another, and low scores with a third, enabling non-expert users to quickly grasp the relative importance of features without interpreting numeric values. This visual encoding maintains measurement precision while dramatically improving ease of operation.
Solution Approach 2:
The patent transforms one-dimensional numeric scores into two-dimensional visual representations by mapping score values to positional arrangements and visual hierarchies in graphical displays. Features with higher contribution scores are positioned more prominently in the visualization, creating an intuitive spatial understanding that preserves numeric precision while enhancing interpretability for non-expert users.
2Adaptability or versatility
If feature contribution scores are standardized across different machine learning models, then adaptability is improved, but manufacturing precision deteriorates
Solution Approach 1:
The patent applies parameter transformation techniques that map feature contribution scores from different machine learning models into a standardized range while preserving relative relationships. By changing the parameter scale and normalization approach, the system achieves cross-model adaptability without sacrificing the precision of individual model predictions, as the transformation maintains the ordinal relationships and relative magnitudes of contribution scores.
Data Source
AI summary
A historical feature contribution score dataset comprising a number of sets of scores generated by machine learning model may be obtained. Additional feature contribution score sets may be materialized such that the size of each additional feature contribution score set is based on a corresponding randomly selected values within a set-size range. A training dataset may be produced that includes feature contribution scores and corresponding classification labels extracted from the historical feature contribution score dataset and the additional feature contribution score sets. The classification labels may indicate an amount that the corresponding feature contribution scores contribute to a prediction of a target feature. A machine learning model may be trained to predict the classification labels using the training dataset. An input feature contribution score set may be applied to the machine learning model to obtain predicted classification labels.