Feasibility study of typhoon disaster economic loss assessment based on random forest

Abstract

Abstract Based on the disaster-causing factors and the actual disaster-causing factors of typhoon disasters, this paper selects 9 assessment indicators, obtains an effective sample set through noise processing, and uses a random forest algorithm to set up a typhoon disaster economic loss assessment model. The model verification and results show that the random forest model has very obvious advantages in both single-level evaluation accuracy and total evaluation accuracy, and can meet the consistency with the actual evaluation standards. The GINI index reduction value obtained through model feedback can be used as an indicator of importance, and the indicator of the number of disaster victims is the most important factor of economic loss.