YAMAMOTO Mana Development of a route selection model for returning home from the “Nagaoka Matsuri Daihanabitaikai”, taking into account information collection methods SANO Kazushi，KATO Teppei Nagaoka City, Niigata Prefecture, hosts the annual “Nagaoka Matsuri Daihanabitaikai” on August 2 and 3, and the concentration of visitors causes serious traffic congestion every year from 9:00 p.m. to midnight, the time when the fireworks display ends. On a particularly congested route, it has been recorded that it took up to two hours or more to pass through a 4.4-km stretch of road. In order to reduce the concentration of demand on Route 8 and alleviate traffic congestion, it is necessary to induce traffic to diversify into detour routes. Soft measures, such as the provision of information in advance, are being taken as a countermeasure, and it is considered necessary to evaluate the means of providing information. Therefore, this study analyzed the attributes of visitors to the “Nagaoka Matsuri Daihanabitaikai” based on the results of a questionnaire survey conducted on the day of the festival, and also constructed a route selection model. The results of the questionnaire indicate that the proportion of repeat visitors who know when to purchase tickets has increased since last year's fireworks festival, when advance purchase of tickets at the venue became necessary. In addition, the percentage of IC use by the number of times of participation in the festival decreased as the number of times of participation increased. This suggests that once a driver is guided to a detour route, he/she is expected to continue to use the route. The responses to the questionnaire were used to estimate the parameters of a route choice behavior model based on a non-aggregate logit model. In this study, we estimated the parameters with the goal of incorporating coefficients related to individual characteristics, such as whether or not information was collected in advance, and tolerance for traffic congestion. Although the likelihood ratio could not be improved with the data used in this study, the t-value and accuracy rate increased compared to models that did not use individual characteristics, confirming an improvement in accuracy. This indicates the possibility that the type of effective policy intervention or device to increase the probability of choosing a detour route may differ depending on the individual characteristics.