
156 Big Data: Techniques and Technologies in Geoinformatics
to select distinct preferred routes and distinct start times. For example, route ranking
may differ across rush hour and non-rush hour and in general across different start
times. However, TD road maps are large and their size may exceed 10
13
items per
year for the 100 million road segments in the United States when associated with
per-minute values for speed or travel time. Thus, industry is using speed proles,
a lossy compression based on the idea of a typical day of a week, as illustrated in
Figure 8.4b, where each (road segment, day of the week) pair is associated with a
time ser ...