There are over 11,500 bus stops in the city of Chicago. For larger
cities such as New York, the number of bus stops may be even
greater. Doing a linear comparison with all the bus stops, buses,
and rail lines is time consuming. Doing a linear comparison took us
over 2 minutes on a HP Laptop with a 4GB RAM and 2.54GHz
Intel Core 2 Duo processor. This is impractical and ineffective in
the real world, since in two minutes, users may transfer from one
transportation mode to another, or become frustrated with the
system. The proposed zip-code based indexing and pruning
approach reduces our feature creation time from over 2 minutes to
below 10 seconds. This can be further improved by using more
sophisticated techniques, such as indexing by R-trees. However,
performance was not a focus of this work.