DETERMINATION OF POTENTIAL FEEDER BUS STOPS FROM SMART CARD DATA AND CAPACITATED CLUSTERING ALGORITHM

Abstract

Most of the literature that deals with feeder bus route design simply assumes the feeder bus stops are already established. The Feeder bus stops are established to connect some known origins to a known destination unlike the conventional bus stop location. In this study smart card data together with geo location of existing bus stops are used to obtain the origin (feeder bus stops) and destination (train station). The first step is to identify the conventional bus stops by isolating smart card users who use these stops to a particular destination. These conventional stops are numerous and hence a demand-constrained clustering method was developed to convert these stops into feeder bus stops. This method was then applied to data extracted from Izmir’s database management system. This methodology was compared to well-known clustering algorithms and it performed reasonably well. This study can serve as a foundation in the application of smart data in the planning of feeder bus routes.