Page 59 - Big Data Analytics for Connected Vehicles and Smart Cities

40 Big Data Analytics for Connected Vehicles and Smart Cities What Is Big Data? 41

agencies. It is typical that the central focus for data collection in a transporta-
tion agency lies in the operations department. As Figure 3.5 illustrates, there is
much more to transportation than operations.
Note that different transportation agencies may have different terminol-
ogy used to describe each of these stages and that Figure 3.5 uses generic terms.
In fact, one of the challenges facing transportation agencies in the light of big
data and analytics will be how to share data and information across the entire
organization in a seamless manner.

Variability
Data analysts and data scientists have an interesting perspective on variability.
They love it. A typical approach to variability is to create averages and summa-
ries to be able to handle it. The data experts see this as a problem because their
perspective is that the value is in the detail and that summaries and averages
remove detail. A good example in transportation would be the average travel
time between traffic signals. While this is a useful measure of performance, it
ignores detail that would be extremely valuable. For example, if a driver trav-
els at 50 mph and then stops at a red light, then travels at onward at 50 mph
and then stops at the next red light, it could be measured that on average the
vehicle traveled at 30 mph. This journey is not differentiated from another
driver who might experience a smooth, steady 30-mph journey through the
corridor, receiving green signals at each intersection. Averaging removes some
important detail. The use of detailed, second-by-second speed profiles for in-
dividual vehicles traveling along a signalized corridor would address this issue.
Such profiles have, in the past, been considered to be unmanageable. It is only
recently that data collection techniques such as probe vehicle data collection
have enabled the acquisition of such data. In the world of big data and analytics
the data is available, and the horsepower to convert it to meaningful informa-
tion is at hand.

Complexity
Data is becoming more complex, and the ability to capture the same data from
more than one data source adds to this complexity. Techniques have been de-
veloped to compare the same data from multiple data sources—this is known as
octagonal sensing. Figure 3.6 provides an information technology–centric view
of all the factors that go along with increasing data complexity.
Figure 3.6 shows a transition from enterprise resource planning (ERP)
that addresses the optimization of assets within an organization to customer
relationship management (CRM) that considers the external interface between
the organization and the customer. This leads on to extensive use of the web,
which is not only driven by big data but can itself generate big data. This migra-
tion from ERP to big data features increasing data variety and complexity but