Abstract
GPS recording devices offer a painless way to collect travel data, but are not directly useful to a standard activity survey. This paper documents one method for linking activities with location data. Based on a small but extended pilot survey, a technique has been developed to estimate the most likely activity at a destination, based on the respondents’ past responses. If destinations and activities were randomly paired, this information would be irrelevant. But the pilot survey also demonstrated that activities are tightly clustered in space.