Wildfire evacuation prediction on Kincade Fire Marshall Fire cross-event (test)
43.3Precision (Stay)FLARE
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| FLARELLM Backend=Claude-3.52025.02 | 43.3 | 86.7 | 57.8 | 96.1 | 74.2 | 83.8 | 76.5 | 70.8 | 79 | |
| FLARELLM Backend=GPT-4o2025.02 | 38.7 | 80 | 52.2 | 94 | 71.2 | 81 | 72.8 | 66.8 | 75.7 | |
| HCM2025.02 | 34.8 | 73.6 | 47.3 | 91.1 | 66.6 | 76.8 | 67.8 | 59.6 | 71 | |
| GPT-4o InferenceLLM Backend=GPT-4o2025.02 | 27.3 | 23.1 | 25 | 82.3 | 85.3 | 83.8 | 73.3 | 54.4 | 72.5 | |
| Logistic Regression2025.02 | 16 | 19 | 17 | 79 | 76 | 78 | 65 | 48 | 66 | |
| Random Forest2025.02 | 0 | 0 | 0 | 80 | 100 | 89 | 80 | 45 | 72 |