Machine Learning Feature Reduction Using Matched Case-Control Subsets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machine learning techniques face inefficiencies in processing large datasets due to the need to consider numerous features, leading to increased computational requirements and reduced model interpretability, while also risking accuracy if irrelevant or redundant features are not adequately filtered.
Innovation Solution
The method involves selecting a reduced subset of features by identifying statistical significance across multiple case-control subsets using propensity score matching and ranking features based on selection scores, ensuring that only the most relevant features are used for model training, thereby reducing computational operations without compromising accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a large number of features are processed in machine learning models, then model accuracy may be improved, but computational requirements and processing time increase significantly
Solution Approach 1:
The patent extracts and removes irrelevant and redundant features from the dataset before model training. By identifying and eliminating features that do not contribute meaningfully to prediction accuracy, the system reduces the feature set to only those that are necessary, thereby decreasing computational requirements while preserving model performance.
Solution Approach 2:
The patent changes the parameter of feature dimensionality by transforming a high-dimensional feature space into a lower-dimensional one through systematic feature selection. This parameter change is achieved by evaluating feature importance metrics and retaining only the top-performing features, thus optimizing the balance between accuracy and processing efficiency.
2Loss of information
If a large number of features are used in machine learning models, then more information is available for prediction, but model interpretability decreases
Solution Approach 1:
The patent extracts and removes irrelevant and redundant features from the dataset before model training. By identifying and eliminating features that do not contribute meaningfully to prediction accuracy, the system reduces the feature set to only those that are necessary, thereby decreasing computational requirements while preserving model performance.
Solution Approach 2:
The patent applies different quality standards to different features by evaluating each feature's contribution to model performance. Features are selectively retained or discarded based on their individual importance metrics, ensuring that only high-quality, informative features are included in the final model, thus maintaining interpretability without sacrificing information availability.
3Productivity
If conventional feature selection methods are used, then feature subset is reduced, but model accuracy is compromised
Solution Approach 1:
The patent performs preliminary feature evaluation and filtering before model training by assessing feature importance metrics across multiple models and iterations. This preliminary action identifies and retains only the most informative features, ensuring that the reduced feature set maintains high predictive accuracy while improving computational efficiency.
Solution Approach 2:
The patent implements a feedback mechanism where feature selection is iteratively refined based on model performance evaluation. Features are selected, models are trained, performance is measured, and the process is repeated with adjustments, allowing the system to converge on an optimal feature subset that maximizes accuracy while minimizing computational requirements.
Data Source
AI summary
A computer system trains a predictive model. A plurality of subsets of features are selected from a dataset comprising a plurality of cases and controls and a plurality of features. Cases and controls are matched to select a plurality of case-control subsets for each subset of features, each case-control subset having similar values for the corresponding subset of features. For each case-control subset, a statistical significance of each feature of the plurality of features absent from the subset of features used to match the case-control subset is identified. A final subset of features is selected based on satisfying a statistical significance of each feature for the plurality of case-control subsets. A predictive model is trained using the final subset of features. Embodiments of the present invention further include a method and program product for training a predictive model in substantially the same manner described above.