Code and data for: Integrating Explainable AI and Multi-Objective Optimization for Wildfire Resource Allocation

This dataset contains the code and data supporting the findings of the manuscript: "Integrating Explainable AI and Multi-Objective Optimization for Wildfire Resource Allocation under Fixed Staffing Constraints: A Decision-Oriented Framework", submitted to Scientific Reports, Submission ID: 10b16a06-52ca-438f-9b5b-ae5d2fddb681. Files included: wildfire_with_label.csv: Dataset of 544 wildfire events in Taiwan (2011–2024) with severity label (MajorFire, binary: 1 = large fire, 0 = contained fire) wildfire_moead.py: XGBoost risk prediction + MOEA/D optimization, reproduces Table 2 and Figures 3–5 wildfire_shap.py: SHAP interpretability analysis, reproduces Table 1 and Figures 1–2 README.md: Full documentation including parameters, expected outputs, and dataset column descriptions Requirements: pip install xgboost scikit-learn pandas numpy matplotlib shap (Python 3.8 or later) All random operations are fixed by np.random.seed(42) and XGBClassifier(random_state=42, scale_pos_weight=2.0).

Expected outputs: Precision=0.622, Recall=0.778, F1=0.691 (threshold=0.20). Key parameters: ALPHA_F1=0.313, BETA_F1=0.354, POP_SIZE=120, N_GEN=250. Contact: joseph_hsu@mail.npust.edu.tw