Partitioning the Time of Day for Bus Schedule Optimization

Time of day partition of bus operating hours is a prerequisite of bus schedule design. Reasonable partition plan is essential to improve the punctuality and level of service. In Istanbul, bus vehicles have been equipped with smart card system. With smart card data passenger boarding time and landing time can be obtained. The number of people between stations in a bus route can be acquired using this boarding and landing time stamps of passengers. In this paper, k-means and fuzzy c-means clustering method will be used and compared to partition the operating day of a bus line in Istanbul into time of day intervals using the bus occupancy rate. Bus occupancy rate will be clustered as low density (very comfortable), normal and overcrowded (disturbing). Membership functions of bus occupancy rates will be based on bus capacity of specified bus route. The resulting clusters could be used in scheduling of bus lines including the frequency and size of buses in different times of day.