Geography Reference
In-Depth Information
because they lived and worked very close to subway stations and they traveled only
in the subways during those days. In addition, the inappropriate usage of the device
(e.g. the GPS device was stored deep in the carrying bag) that make it difficult to
receive signals also accounted for some of the discrepancies.
We have also separately calculated the successful rate for the 1-min data and that
for the 3-min data. The results show that the successful rate of the 1-min data was
84 %, whilst that of the 3-min data was 82 %. Though that for the 1-min data is
higher than that for the 3-min data, but the difference is very marginal. However,
the 3-min data saved lots of storage space, battery life and preprocessing efforts.
13.5.1.4
Identifying Possible Trip Purposes
To identify the possible trip purposes associated with trip ends, in addition to the
POI information, individuals' basic information like the addresses of home, school
or work place, places that they often visit are also used. Based on the addresses that
respondents filled in the questionnaire, we made use of Google Earth and ArcMap
to obtain the exact positions of these places in terms of latitude and longitude
coordinates. A separate layer of the positions of home, workplace and often visited
places was established for each participant. Once this was done, we then associated
trip ends data with personal basic data and the land-use types nearby to identify
possible trip purposes using buffer analysis. In the earlier years of GPS-based travel
surveys, when the accuracy of positioning is concerned, Wolf et al. (
2004
)define
that trip ends within 200 m from the home location were assigned with the activity
purpose 'home', and other point of interests within the distance of 300 from each
ends were derived as possible purposes. Years later, with the improvement of GPS
technologies, Bohte and Maat (
2008
) reduce the area of buffer zone and argued
that the area of 'home' and 'work' should be larger since the location of which
would be visited frequently (i.e. if distance between trip end and POI <50 m,
then category
D
category POI ('shopping', 'recreation', 'culture', 'medical', 'kids'
or 'railway station', if distance between home/work and trip end < 100 m, then
category
D
'home'/'work'). Similarly, Moiseeva et al. (
2010
) also set the radius of
100 m for the personal spatial data and 50 m for the land use data in the region of
interest. Consequently, taking the previous experience and positioning accuracy into
consideration, buffer zones in this case study were defined as within a distance of
100 m or 50 m respectively from home/work address and other frequently visited
places/point of interest. Finally, Land use features which fall into the buffer zones
of trip ends are regarded as possible purposes.
13.5.2
Operationalizing and Calibrating the Algorithm
Based on the above discussions, five variables including land use types of trip ends
(in terms of possible trip purposes), trip durations, the starting time of the trips,
age of the respondents and day of the week are used to encode the chromosome.
Search WWH ::
Custom Search