Day on day dependencies in travel: First results using ARIMA modelling

Abstract

In this paper, we have used a technique for the analysis of a 42-day continuos record of travel behaviour data that has not been used before for travel diary and transport planning data. The travel diary data of the MobiDrive dataset has been examined from a different perspective with the use of a specific class of stochastic models called ARIMA (Box and Jenkins, 1976), which have been used to extrapolate day on day correlations and weekly patterns from the data col- lected. An ARIMA modelling approach has been conducted on an individual basis for each subject of the study and for a set of variables related to trips, distances and durations of daily activities. This approach has allowed a preliminary classification of the subjects of the Mobi- Drive study into different model categories, and can be used as a starting point or a more de- tailed categorisation for the application of different analysis techniques.