Feature Selection via Sub-Population Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Feature selection techniques in machine learning often fail to identify informative features, leading to reduced accuracy and misallocation of processing resources in predictive models.
Innovation Solution
A computer system employs sub-population analysis to select additional features by matching cases and controls with similar values, using propensity score matching and statistical significance testing to enhance feature selection and improve predictive model performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional feature selection techniques are used, then processing efficiency is maintained, but predictive model accuracy deteriorates due to failure to identify informative features
Solution Approach 1:
The dataset is segmented into sub-populations based on similarity of feature values. By dividing the data into distinct sub-populations and performing feature selection within each segment, the method identifies features that are informative within specific contexts, thereby improving predictive accuracy without requiring exhaustive analysis of the entire dataset.
Solution Approach 2:
Propensity score matching is used as an intermediary mechanism to identify and match cases with similar feature profiles. This intermediary step enables the systematic identification of informative features by comparing outcomes within matched sub-populations, improving accuracy while maintaining computational feasibility through structured comparison.
2Measurement precision
If comprehensive feature selection is performed on the entire dataset, then all potentially informative features are identified, but processing resources are misallocated and complexity increases
Solution Approach 1:
The feature selection process applies local quality by identifying features that are informative within specific sub-populations rather than requiring all features to be informative across the entire dataset. This localized approach reduces the number of features that need to be evaluated globally, simplifying the overall process while maintaining high accuracy for predicting outcomes within each sub-population.
3Measurement precision
If sub-population analysis with propensity score matching is applied, then additional informative features are identified, but processing time and computational resources increase
Solution Approach 1:
The method applies partial action by performing feature selection on a subset of matched cases within sub-populations rather than analyzing the entire dataset. By selecting representative samples from each sub-population and performing feature selection on these smaller, matched sets, the approach identifies informative features effectively while reducing overall processing time and computational resource requirements.
Data Source
AI summary
A computer system selects features of a dataset for predictive modeling. A first set of features that are relevant to outcome are selected from a dataset comprising a plurality of cases and controls. A subset of cases and controls having similar values for the first set of features is identified. The subset is analyzed to select a set of additional features relevant to outcome. A first and second predictive model are evaluated to determine that the second predictive model more accurately predicts outcome, wherein the first predictive model is based on the first set of features and the second predictive model is based on the first set of features and the additional features. The second predictive model is utilized to predict outcomes. Embodiments of the present invention further include a method and program product for selecting features of a dataset for predictive modeling in substantially the same manner described above.