The modeling of dwelling time of buses at bus stop

Abstract

This research is aimed to build generic dwelling time modeling for BRT in Indonesia. Dwelling time was counted from the time of the bus entering to leaving the stop, including additional passenger service time. The observations were captured on 172 services period for 6 hours each at 6 heterogeneous bus stop using a video camera. The cameras were mounted inside the bus stop. It was located across stop's gate. This cameras' spot sets to capture the bus and passenger movement time to time. The clustering process used Pearson's correlation to differentiate of several data groups. The groups are global data, passenger by direction and additional data services. The Pearson's correlation value shows best on data that is divided into three categories, which are alighting time, departure time and additional service time. Each category is fitted into several regression models candidates, such as linear, polynomial, power, and logarithmic. A single linear regression model was found valid for representing bus dwelling time for each category. There the dwelling time model is a compound of three categorical model.