Escalating wildfire threats in the western United States have created a critical gap between large-scale emergency alerts and actionable household preparedness. This research presents the Wildfire Readiness Assistant (WiRA), an AI-driven decision support system designed to bridge this gap by providing regionally localized readiness guidance. Utilizing a Logistic Regression classifier trained on CAL FIRE perimeter data and historical weather archives, the system prioritizes interpretability and safety-critical performance. Validated on a statewide 2022 California fire-season dataset, the model achieves a 94.1% recall rate for the binary classification of elevated- versus low-risk conditions (evaluated on the binary Elevated vs. Low classification task, as no High-risk observations were present in the 2022 dataset), ensuring that potential threats are identified while accepting a precision trade-off appropriate for recall-first public-safety decision support. The study explicitly addresses data reliability through strict temporal consistency validation and implements a "Human/Official Boundary" architecture. This approach demonstrates how machine learning can effectively augment household decision-making in public safety contexts without usurping official emergency directives.

Keywords

Wildfire Risk Assessment, Machine Learning, Decision Support Systems, Emergency Preparedness, Public Safety.