Choice Set Generation from Rule-based GPS Data Processing and Analysis of Factors to Choose Grocery Store

Abstract

In this study, choice set for the application of disaggregate logit model is attempted to generate from GPS data set, which is obtained by logging to 3,500 persons for a week in Zurich. The concept of choice set generation is on the basis of rule-based simple method. Three rules for GPS processing are proposed, such as; (i)extraction of grocery shopping trips, (ii)identification of x-y geo-code for activity places, and (iii)identification of places for individual’s home and work/school, in order to apply for disaggregate choice model. Choice set is prepared for each transport mode such as automobile, walk, and public transport. The results by using proposed rules are compared with microcensus data in terms of several aspects of trip characteristics and confirming the validity of proposed rule. Using generated choice set, grocery store choice model is developed. Estimated parameters are received as reasonable signs and statistics. From these results, we can confirm the proposed method for choice set generation is reasonable enough for the application of disaggregate logit model.