ABSTRACT

Time series models have been the focus of considerable research and development in recent years in many disciplines, including transportation. This interest stems from the insights that are gained when observing and analyzing the behavior of a variable over time with a time series being a sequence of observations arranged by their time of outcome. Modeling and forecasting a number of macroscopic traffic variables such as traffic flows, speeds, and vehicle occupancies are common transportation-related time series problems. This chapter explores important elements of time-series data including trend, seasonal, and random components and various filtering strategies that can be undertaken with the data. Concepts of stationarity and dependence are presented and discussed along with elements of time series regression including serial correlation, dynamic dependence, volatility, and causality. The chapter underscores the great importance of time series approaches in transportation data analysis.