A Typological Approach to Individual Urban Travel Behavior Prediction

Abstract

With the use of cluster analysis a sample of 1018 residents of the San Francisco Bay Area was classified into eleven types of urban residents on the basis of overall similarity of personal and environmental characteristics, and independently into nine types of travel behavior. The relations between the two typologies, and the comparative travel behavior of the types of urban resident were investigated in an attempt to gain insight into the determinants of urban travel. The probability of the correct assignment of a travel behaviour type to an urban resident type was of the order of 0·30. Monte Carlo simulation methods were used to test empirically whether the value of a given travel behavior characteristic for a given urban resident type can be assumed to be higher (or lower) than the value in the general population, thus testing the predictability of the travel behavior of the various urban resident types. Conversely, the prediction of the urban traveler's personal characteristics given his travel behavior type was also evaluated. This typological approach made the prediction of the usage of the San Francisco Bay Area rapid transit system (for going to work, for going to shop, for going out for leisure, or for some other purpose) possible, in about 15% of the cases, from the knowledge of the urban resident type and, in about 25% of the cases, from the knowledge of the travel behavior type.