Abstract: To address the challenge of emergency supplies demand forecasting in the early stages of typhoon disasters, where data are often incomplete and traditional methods are highly dependent on prior assumptions, this paper innovatively optimizes a case-based reasoning forecasting model. First, the entropy weight method and principal component analysis are integrated to determine the weights of disaster characteristic attributes, and case similarity is calculated through weighted distance metrics. Second, a random forest regression model is introduced, which incorporates demographic and socioeconomic indicators, to construct an event-outcome-based case revision mechanism. Furthermore, a multi-case weighted comprehensive reasoning approach is adopted to predict disaster consequences, and the total demand is estimated by integrating established material supply standards. The results demonstrate that the proposed method can effectively forecast emergency material needs under data-scarce conditions. It is operationally straightforward and highly adaptable, providing reliable support for emergency decision-making and material allocation during the initial disaster response phase.

Key words: typhoon disaster, material demand forecasting, case-based reasoning, case revision mechanism