Model Analysis
View and manage available machine learning models for voter propensity scoring.
Active Model
Enhanced Australian Treaty Referendum Model v2.0
Active
Available Models
Enhanced Australian Treaty Referendum Model v2.0
activeAdvanced machine learning model with sophisticated feature engineering for Australian treaty referendum voter propensity prediction. Incorporates 18 engineered features including wealth stability, home stability, age-education interactions, and state-specific engagement patterns. Uses Gradient Boosting with hyperparameter optimization and cross-validation for superior performance.
Version: 2.0.0
Accuracy: 94.6%
Updated: 2024-01-20
General Election Model v0.9
inactiveExperimental model for general election voting patterns
Version: 0.9.2
Accuracy: 82.0%
Updated: 2024-01-10
Enhanced Model Features
- 18 engineered features (9 original + 9 new)
- Advanced feature engineering (wealth_stability, home_stability)
- Cross-validation with hyperparameter tuning
- Multiple algorithms (XGBoost, Random Forest, Gradient Boosting)
- State-specific engagement patterns
- ONNX format for fast web inference
Enhanced Performance Metrics
ROC AUC Score:0.946
Accuracy:86.9%
Precision:88.3%
Recall:87.9%
F1 Score:88.1%
Top Feature Importance
Wealth Stability:28.9%
Home Stability:24.4%
Age-Education Score:21.6%
Education-Income Score:13.4%
Income Numeric:3.2%
Other Features:8.5%