Time Series Diagnostics & What-if scenario analysis
Intervention-Aware Modeling
Understanding how localized changes in one variable affect others in multivariate time series is essential for diagnostics and decision-making. Our framework combines temporal decomposition with frequency-domain feature correlation modeling to enable realistic "what-if" scenario analysis.
Key Innovations
- Realistic propagation across correlated variables
- Target feature prediction and simulation analysis for updated features
Applications
Time Series Forecasting · Decision Support Systems
Request Information
What-If Feature Simulation: Analyze how modifying an input feature (e.g., DEWP) impacts the target prediction — helping operators understand causal dependencies.
[figures/adjusted_prediction.png]
Effect of modifying an input feature (DEWP) on the model's prediction. "Adjusted Prediction" illustrates how the target output responds to the modification, highlighting the model's sensitivity to feature-level changes.
Time Series
What-if Analysis