Scenario Simulator
Pagsubok at paghahambing ng napatunayang makasaysayang sitwasyon ng sakuna
Lalawigan ng Bataan, PH
Thu, 22 Oct 2026
Ligtas na pag-sign in
What-if tool

Scenario Simulator

Try out a hypothetical disaster situation and see how the Gradient Boosting model, trained on the latest completed AutoML run, would classify its damage risk.

ML what-if classification: This estimates a damage-risk class only. It does not predict when a disaster will occur or estimate an exact peso damage amount.
Run a scenario

Inputs and Result

Fill in the situation on the left; the model's classification appears on the right

Scenario Inputs
Values available before final damage is known
Optional assistance inputs
Scenario Result
Binary disaster damage-risk classification
Ready to simulate Complete the scenario inputs and select “Run Risk Scenario.”
ModelGradient Boosting
Final validation F10.6667
Final validation records8
Training source30 recorded events
Model context

How the Model Sees Each Event Type

Grounds the scenario above in real historical patterns

Historical Model Pattern by Event Type
Average out-of-fold high-risk probability from 30 recorded events; hover for exact values
Model basis and usage limitations

The simulator uses Gradient Boosting with tuned settings: learning rate 0.03, max depth 1, n estimators 50, random state 42. The deployment model was fitted to all 30 historical records only after model selection and validation.

Damage columns are not used as predictors, preventing direct target leakage. The final untouched holdout contained 8 events and produced F1 = 0.6667, so the result supports planning and comparison but must not be presented as a guaranteed future outcome.