Predictive Model Segmentation for Imbalanced Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current classification techniques struggle with imbalanced datasets, particularly when positive cases are scarce, leading to high false positive rates and inaccurate predictions due to incorrect pattern learning.
Innovation Solution
Generate multiple predictive models based on individual positive samples from an imbalanced dataset, identify influential parameter groups, and combine these models to create a final predictive model that classifies samples accurately.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional classification models are used on imbalanced datasets, then the model can be trained with available data, but the positive case detection accuracy deteriorates due to insufficient positive samples
Solution Approach 1:
The patent creates synthetic copies of positive samples by generating multiple predictive models, each focused on different aspects of positive case detection. These copied models are then aggregated to form a comprehensive detection system that overcomes the limitation of insufficient positive samples
Solution Approach 2:
The patent segments the classification task into multiple specialized predictive models, where each model focuses on specific parameter groups influential for different positive samples. This segmentation allows each model to specialize in detecting particular patterns, improving overall positive case detection accuracy
2Reliability
If traditional models are trained on imbalanced datasets, then training can proceed with existing data distribution, but false positive rates increase due to incorrect pattern learning
Solution Approach 1:
The patent applies local quality by identifying and focusing on specific influential parameter groups for each positive sample rather than treating all parameters uniformly. Each predictive model is tailored to specific parameter combinations, improving pattern recognition accuracy and reducing false positives
Solution Approach 2:
The patent performs preliminary analysis to identify influential parameter groups before training the predictive models. This preliminary action ensures that models are trained on the most relevant features from the beginning, preventing incorrect pattern learning and reducing false positives
Data Source
AI summary
Methods, systems, and computer program products perform classification based on an imbalanced dataset. In a method, machine learning models are generated based on positive samples included in an imbalanced dataset. An amount of the positive samples is less than an amount of negative samples that are included in the imbalanced dataset. Each sample in the positive and negative samples includes parameters. Multiple influential parameter groups are identified from the parameters for the positive samples, respectively. A final predictive model is determined based on the machine learning models and the multiple influential parameter groups. The final predictive model is used for classifying a sample as a positive type or a negative type.