Typhoons, as highly influential meteorological disasters, cause extensive damage to both property and lives. The accurate and efficient prediction of typhoon disaster losses (TDL) is crucial for pre-disaster risk management, emergency response strategies and post-disaster loss assessment. While existing studies typically reveal linear relationships between TDL and its influencing factors, the focus need to be shifted toward developing non-linear models that can capture the complexities of these relationships. To this end, this study leverages historical TDL data at the county level from 2001 to 2023 in Zhejiang Province, China. Combining RF, XGBoost, LightGBM, MLR models and SHAP values, based on meteorological, environmental and economic data, this study develops interpretable TDL prediction models. The results demonstrate that the XGBoost model performs best with a R 2 of 0.7060. The most important influencing factors are accumulated precipitation, the nighttime light index, maximum daily extreme wind speed (max EWS) and maximum daily maximum wind speed (max MWS). Meteorological factors, especially wind speed and atmospheric pressure, were found to be the most significant contributors to TDL prediction.
Wu et al. (Sun,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: