| dc.description.abstract | Origin-Destination surveys, which are regularly conducted in many cities (to
calibrate transport models), contain indirect information on individual time use that can be
recovered through the declared trip purpose. Although this data source is very rich, it has
two limitations for the calibration of time use models: the level of disaggregation regarding
time use is constrained by the definition of trip purposes, and the information gathered on
different time periods is usually obtained from different individuals. In this paper we
propose a new method to overcome the second limitation, transforming the original daily
observations into individual-weeks. For every working day observation we build Saturday
and Sunday ‘‘twins’’ as a convex combination of observed weekend individuals such that
the distance between the attributes of the working day individual and the synthetic twin is
minimized. We applied this procedure to the Santiago OD survey, and generated a database
of weekly observations particularly rich for model calibration and segmentation. | |
