Application of gamma models for off-peak dwell time in urban transit routes: a case of study in the Greater Metropolitan Area, Costa Rica

The dwell time at bus stops is affected by factors such as the number of passengers, the method of payment, the configuration of the buses, among others. Therefore, to find the necessary balance in public transportation, where the comfort of the users and the operational needs of transit agencies are the best, it is necessary to make an efficient route design and optimize the methods of payment and boarding. Generally, the analysis of this problem is carried out through multivariable regressions, either linear or non-linear, where the variables that are taken into account depend on the focus of each study. In this case, a generalized linear model is used, using the gamma distribution. The application of generalized linear models using the gamma distribution allowed the modeling of the phenomenon studied and determine the factors that significantly affect the stoppage time. It was determined that, when comparing the same type of bus and on the same route, the main factors that directly affect the bus stop time are the number of users, the method of payment and the configuration of the buses. By generating a general model, including all available data, it is possible to evaluate some of the parameters responsible for dwelling time variations. This allows to establish the effect when implementing the use of buses with low floor in this public service, obtaining a decrease of approximately five seconds in each stop that the vehicle makes along its route.