Reliable Short-Term Traffic Flow Forecasting for Urban Roads Multivariate Generalized Autoregressive Conditional Heteroscedasticity Approach

• Published in: Transportation research record Vol. 2343; no. 1; pp. 77 - 85
• Main Authors: Xia, Jingxin, Nie, Qinghui, Huang, Wei, Qian, Zhendong
• Format: Journal Article
• Language: English
• Published: Los Angeles, CA SAGE Publications 2013
• Subjects: Forecasting; Mathematical analysis; Mathematical models; Roads; Traffic engineering; Traffic flow; Transportation; Vectors (mathematics)
• ISSN: 0361-1981; 2169-4052
• DOI: 10.3141/2343-10

Short-term traffic flow forecasting has an essential role in advanced traveler information systems, route guidance systems, and proactive traffic signal control systems. Numerous univariate and multivariate models have been presented on both traffic flow level forecasting and traffic flow variance forecasting. However, few studies have incorporated the relationship between traffic parameters (such as volume and speed) into development of the traffic flow forecasting model. It is well known that there are inherent relationships between traffic parameters and that heteroscedasticity exists in traffic flow series. On this basis a vector autoregressive plus multivariate generalized autoregressive conditional heteroscedasticity (GARCH) method is proposed for reliable short-term traffic flow forecasting for urban roads. The vector autoregressive model is used as the mean equation of the multivariate GARCH model for modeling traffic flow levels, and the multivariate GARCH model is used to model the conditional traffic flow variances. Actual traffic volume and speed data are used to validate and evaluate the proposed method. Results show that the proposed method can generate workable performance in forecasting accuracy and forecast confidence intervals.