An Online Change-Point-Based Model for Traffic Parameter Prediction

This paper develops a method for predicting traffic parameters under abrupt changes based on change point models. Traffic parameters such as speed, flow, and density are subject to shifts because of weather, accidents, driving characteristics, etc. An intuitive approach of employing the hidden Marko...

Full description

Saved in:
Bibliographic Details
Published inIEEE transactions on intelligent transportation systems Vol. 14; no. 3; pp. 1360 - 1369
Main Authors Comert, Gurcan, Bezuglov, Anton
Format Journal Article
LanguageEnglish
Published IEEE 01.09.2013
Subjects
Adaptation models
Change point models
Data models
Forecasting
hidden Markov model (HMM)
Hidden Markov models
Numerical models
Prediction algorithms
Predictive models
time-series autoregressive integrated moving average (ARIMA)
traffic prediction
Online AccessGet full text

Cover

More Information
Summary:This paper develops a method for predicting traffic parameters under abrupt changes based on change point models. Traffic parameters such as speed, flow, and density are subject to shifts because of weather, accidents, driving characteristics, etc. An intuitive approach of employing the hidden Markov model (HMM) and the expectation-maximization (EM) algorithm as change point models at these shifts and accordingly adapting the autoregressive-integrated-moving-average (ARIMA) forecasting model is formulated. The model is fitted and tested using publicly available 1993 I-880 loop data. It is compared with basic and mean updating forecasting models. Detailed numerical experiments are given on several days of data to show the impact of using change point models for adaptive forecasting models.
ISSN:1524-9050
1558-0016
DOI:10.1109/TITS.2013.2260540