Human travel behaviour can be considered as a sequence of activities, which are characterised by a number of different dimensions. As individual behaviour is limited by personal or spatial constraints it is crucial to take the different dimensions into consideration as well as the sequential character of behaviour. Traditionally the comparison of activity pattern has been conducted with measures which are based on geometrical (Euclidean) distance measures. These measures do not take topics like the correct order of activities or the different length of patterns into account. In order to take these factors into consideration Wilson (1998) proposed to adopt the method of sequence alignment from the field of molecular biology (Sankoff and Kruskal, 1983) for travel behaviour research. Sequence alignment is based on the Levenshtein distance, which is the smallest number of changes needed to equalise two sequences. Due to the lack of longitudinal data most comparisons between activity patterns have been performed between different persons. This paper focuses on the intrapersonal comparison of activity patterns, especially highlighting their variability and stability. The results presented are based on data from a six week travel diary conducted in the study "Mobidrive" (Axhausen, Zimmermann, Schonfelder, Rindsfuser and Haupt, 2000) and calculated with the software program ClustalG by Wilson, Harvey and Thompson (1999), which is a promising implementation of the sequence alignment method. Additionally, some disadvantages of the method are discussed with special attention given to some proposals by Arenze and Timmermanns (2000) to improve the sequence alignment method. The results are also compared to some simpler methods for pattern comparisons suggested by Huff and Hanson (1986), Pas (1983) and Jones and Clarke (1988).
Abstract