1-V_pm

[66] Estimation of Dynamic Activity Path Choice Model Correcting the Two Biases of Tsunami Evacuation Data

○Takuya Iizuka¹, Eiji Hato¹ (1. University of Tokyo.)

Keywords:Evacuation, Activity path choice, Parameter Estimation, Bias correction

In a disaster management, methods for accurately predicting evacuation behavior are needed. We model a series of dynamic decision makings from the strikes of disasters to the completion of evacuations. Evacuation behavior data have difficulties in parameter estimation; 1) behavioral data from past disasters do not include responses of victims, 2) survey data for virtual disasters contain normalcy bias. In this paper, we model the evacuation behavior by an activity path choice model and propose a method to correct the bias contained in evacuation data in parameter estimation. This method allows to evaluate the staying time at the start or stopping point of evacuation routes. Numerical experiments and case studies showed the effectiveness of this method.