UC Irvine Institute of Transportation Studies

This paper presents a methodology for the analysis of activity patterns based on a classification procedure in which the set of measurements that define human movement is represented by an N-dimensional pattern vector. Transformation techniques are then applied to the pattern vectors to develop a taxonomy for the pattern space. Subsequent inversion of the transformed patterns yields representative activity patterns and leads to attendent transformation of the results of analysis to the real world. Pattern recognition theory is demonstrated to be an effective means by which complex activity/travel patterns can be transformed into a structurally simpler space for purposes of analysis.