01626029

Component

The primary objective of this paper is to develop hybrid short-term bus arrival time prediction models using mixed multi-route bus arrival and departure time data. The proposed hybrid models, consisting of a group of running time prediction submodels and dwell time prediction submodels, aim to minimize arrival time prediction errors by extracting information from multiple routes. To mix and fully utilize the bus arrival and departure time data from multiple routes, three weighted average times are introduced as new model inputs. The widely used prediction algorithm, support vector machine (SVM), is picked up according to its good performance when working with data from multiple routes. Bus arrival and departure time data at 12 stops from Yichun, China, covering 16 bus routes are collected to train and evaluate the proposed models. The performance results demonstrate that the introduction of mixed multi-route data and the hybrid models can significantly improve the prediction accuracy. The best running time prediction and dwell time prediction submodels are model R1 and D7 respectively, while the optimal hybrid bus arrival time prediction model is not the combination of R1+D7, but R1+D1. The average prediction error reduction of model R1+D1 over the conventional model is around 14%. Discussion about model performance illustrates that the prediction accuracy improvement of R1+D1 is larger during peak hour and with heavy mixed traffic condition than during off-peak hour and with moderate mixed traffic condition.

This paper was sponsored by TRB committee AP050 Standing Committee on Bus Transit Systems. Alternate title: Hybrid Short-Term Bus Arrival Time Prediction Models Based on Mixed Multiroute Arrival and Departure Time Data

01618707

17-01869

English

16p

2017

Transportation Research Board 96th Annual Meeting

Digital/other

Figures; References; Tables

Algorithms; Bus transit operations; Dwell time; Mathematical prediction; Peak hour traffic; Running time; Vehicle mix

Arrival time (Bus transit); Support vector machines

Yichun (China)

Highways; Operations and Traffic Management; Planning and Forecasting; Public Transportation

Transportation Research Board Annual Meeting 2017 Paper #17-01869

TRIS, TRB, ATRI

Dec 8 2016 10:39AM