A New Strategy For The Diagnosis Of The Bus Headways Using AVL Data

Abstract

Different properties of transit systems determine the quality of service of a public transportation system and reliability is a significant one. Although there is no single concept of reliability, regularity of the headways of the buses is an important measure. Speed and headway information of buses can be easily extracted from Automatic Vehicle Location (AVL) data with the explained method. This study focuses on a novel measurement using kurtosis of distribution of distance headways among buses along a transit route, which is proposed for identifying irregularities from AVL data of buses during public transportation operations. Kurtosis measurement on headway distribution of 5-minute data can provide online problem detection without using historical data each time step. 20 million trajectory data points of buses in Istanbul, Turkey are analyzed. It was determined that for values above the threshold of kurtosis measure, there was always an anomaly in the system. Transit operators can follow the proposed analysis of one-week AVL data in order to calibrate a kurtosis threshold value for their network. The findings of this research are valuable for both policy makers and operators in cities having large-scale bus networks. This brief methodology can be easily followed to create a tool that signals operators for problems without investigating each route perpetually.