Estimation of Posterior Distribution for Event Effect on Theme Park Problem

Abstract

For theme park operators, the implementation of temporal event in theme park is expected to change visitor behavior and ease waiting times for attractions. It is difficult to directly measure the event effect itself, however, it seems to be useful for the theme park business to estimate the event effect from the observation data. In this paper, we propose a method to estimate the event effect under the situation that the temporal event is implemented in the theme park using multi-agent simulation (MAS) and particle filter. Numerical experiments which use artificially generated waiting time from the theme park MAS as observation data show that the proposed method can estimate the posterior distribution of event effect. Filter distribution and smoothing distribution were estimated as a posterior distribution, and each characteristic is shown. We confirmed that the estimation accuracy of the posterior distribution was improved by considering the information of the time schedule in which the event was implemented in the estimation.