Predicting the Peaking of Holiday Traffic (Victoria Day Example)

In developed and fast developing countries, rising standards of living and the trend towards shorter working hours have significantly changed people's traveling behaviour. Literature reported that, during holiday periods, there were usually substantial increases in traffic volumes on highways. A clear understanding of the holiday traffic characteristics and a reasonable prediction of the upcoming high traffic volumes would significantly benefit both researchers and practitioners who are responsible for the planning, design, operation, and management of transportation networks. However, research focusing on holiday traffic has been minimal to date. The existing traffic prediction methods mainly aim on the flow during regular times. This paper is intended to first investigate the variation characteristics of holiday traffic. Then, the potential prediction errors using popular predicting methods such as time series are discussed. At last, a non-parametric method is proposed to predict the holiday traffic for different types of highways. On the basis of traffic volume data from major highways in Alberta, Canada, it is found that the performances of the proposed method are consistent and reasonable for different holiday periods and various types of roads.