Development of Travel Frequency Forecasting Model for the Elderly

In order to maintain an independent life in remote rural areas, it is necessary to provide accessible transit services. When the service is introduced or improved, profitability that is derived from operation costs and fare receipts should be considered. This paper aims to develop a travel frequency forecasting model by transit services for the elderly. The proposed model predicts individual travel frequency per month after introducing new bus alternatives or improving bus service levels. A discrete continuous model is estimated to forecast not only bus choice probability but also bus travel frequency per month simultaneously. Moreover, it is required to predict travel demand hidden by low levels of services (i.e. latent travel demand) in order to evaluate transit services precisely. To this end, a non-compensatory utility function is employed to represent conjunctive choice strategies in which marginal utilities of decision makers should be sensitive around a threshold. This model can be estimated based on utility maximization theory. A questionnaire survey was carried out in a mountainous area where several types of transit already exist. The survey consisted of questions regarding individual socio-economic characteristics, travel frequency by transit, and boundaries (i.e. minimum levels) of services that users can accept. The boundaries are used as thresholds of the non-compensatory model. Results of model estimation showed that the non-compensatory model has more accurate explanatory power of the elderly's travel choices than the conventional linear utility model. Furthermore, a discrete continuous model can be applied to a travel frequency forecasting model for transit services. It is found that the proposed model can be effective to make transit service planning efficient and to evaluate the effects of improving levels of transit service.