Public Transit Origin–Destination Matrix Determination Using Historical Tracking of Automated Databases

Great efforts have been made and significant capital has been invested in obtaining, storing, and processing public transportation automatic data. Although public transportation systems require numerous data inputs for proper operation, most of this automatic data have not been utilized. This study focused on estimating the public transit origin–destination (OD) matrix using automated fare collection and automatic vehicle location data. Usually, OD matrices are calculated using the classic four-stage model (trip generation, distribution, modal split, and assignment). The OD estimation in this study is based on a disaggregated model and does not follow the classic model order. The data were obtained from the entry-only transaction system in Mashhad, Iran. A trip chain heuristic was developed to determine passengers’ alighting stops/stations. Then, another heuristic was developed to determine the passengers’ destination zones. The results for OD estimation yielded a 253 by 253 matrix, along with trip production and attraction for each zone. Some important performance measures, such as the transfer rate, were also calculated. The results were verified by extensive manual data collection and showed that the proposed method has an average root mean square error per line equal to 3.56 persons per hour. Based on the results, this study shows that the proposed method offers a less labor-intensive and more accurate way to estimate the OD matrix compared to traditional methods of determining the origin–destination matrix.