Training rows
--Inputs
Five preserved model inputs drive both predicted outputs under a commercial-project baseline.
Live forecast
--
Predicted viability score
Predicted construction cost
--
Property rows
--Cost model R2
--Viability model R2
--Baseline viability
--Viability delta
--Baseline cost
--Cost delta
--Live Scenario Impact
Baseline uses the restored model defaults. Each bar shows the isolated effect of one input change against that baseline.
Feature Importance
Compare what matters most for each Random Forest target.
Random Forest Fit
These plots compare model predictions against held-out test data from the restored workflow.
Residuals and Relationships
Inspect model error and the dataset patterns behind the preserved workflow.
Correlation Map
Correlation does not define causality, but it does show how the restored training data is structured.
Model Notes
Loading model metadata...