Bus passenger Origin-Destination Matrix estimation using Automated Data Collection systems

Abstract

Automatic Data Collection (ADC systems can enhance the ability of transit agencies to obtain useful planning information that was previously too expensive to obtain. This thesis documents the development of an algorithm to estimate a Bus Passenger Trip Origin-Destination Matrix (OD Matrix) based on ADC system archived data including Automated Fare Collection (AFC data, Automatic Passenger Count (APC data, and Automatic Vehicle Location (AVL) data. This algorithm consists of three steps: data preparation, estimation of a Single Route OD Matrices for all routes, and estimation of a Network Level OD Matrix using transfer flow information. The single route OD matrix estimation requires the development of a "seed" matrix derived primarily from AFC data and the "marginal" control totals, i.e. the boarding and alighting counts, derived primarily from APC data. Both Iterative Proportional Fitting (IPF) and Maximum Likelihood Estimation (LE) techniques are used to estimate the single route OD matrices based on seed matrices and marginal values, and the results from these two techniques are compared. For the network level OD matrix estimation, we estimate the transfer flows by considering the consecutive transactions from AFC data.

(cont.) The resulting network level OD matrix is provided at a route segment level (3-5 combined stops) of detail. This OD estimation algorithm is illustrated by its application to a selected corridor of the Chicago Transit Authority (CTA) bus network This OD estimation algorithm can be practically applied to a full-size bus network It is recommended that the MLE method be used to estimate the single route OD matrices and the proportional distribution method be used to estimate the transfer flow OD matrix.