Time series decomposition model for traffic flow forecasting in urban midblock sections

Abstract

Predicting the vehicular flow or traffic volume is an integral part in Intelligent Transportation Systems (ITS) applications. The use of time series analysis techniques such as Box-Jenkins Autoregressive Integrated Moving average (ARIMA) as attempted in many studies generally requires large quantity of data for model development. In case if sufficient data are not available, it may not be possible to use ARIMA for traffic volume prediction. To overcome this problem, other time series methods such as multiplicative decomposition can be used as they are easier to understand as well as implement when compared to ARIMA models and the present study is an attempt in this direction. A midblock section in a busy arterial road in Vellore, Tamilnadu was considered and limited traffic volume data from two consecutive days collected between 7 am and 11 am was used for model development using multiplicative decomposition technique. The results are promising and it was found that the mean absolute percentage error (MAPE) between observed and predicted volume falls between 9 and 16, which is acceptable in many ITS applications.