2018 Volume 74 Issue 5 Pages I_201-I_208
Most existing studies on residential choice of disaster victims are based on standard multivariate analysis and advanced method can give us useful insights. This study employs two advanced methods, group-based discrete choice model and decision tree, for analyzing the residential preference of households in Mashiki temporary housings following the 2016 Kumamoto earthquake. Group-based discrete choice model reveals the effect of each household member's situation on residential preference (e.g. possibility of driving a car). Decision tree analysis on single households who were homeowners demonstrates the combination of conditions to explain the residential preference. For example, single households who have been single before earthquake, drivers, and in their 50s are less likely to rebuild their home.