Page 219 - Big Data Analytics for Connected Vehicles and Smart Cities

200 Big Data Analytics for Connected Vehicles and Smart Cities Practical Applications and Concepts for Transportation Data Analytics 201

• Reduce crash frequency, severity, and the likelihood of secondary crashes
by reducing the speed of vehicles as they approach an incident, traffic
queue, or stoppage;
• Stabilize and smooth traffic flows (consistent speeds within lanes and
between lanes).

VSL has been implemented in other cities and studies show the following.

• VSL allows travel at a slower, but more consistent speed, as opposed to
the constant stop-and-go traffic typical of rush hour conditions.
• By regulating traffic speed, VSL also helps reduce rear-end and lane-
change collisions associated with sudden stops at the back of congested
areas.
• This more consistent speed improves safety, saves motorists gas, and
lessens harmful emissions from idling in stopped traffic.

Approach
The overall approach to the work involves a before-and-after analysis of traffic
speeds at one-minute increments on segments that comprise the study area. To
understand the underlying patterns of traffic variation along freeways in the
study area, the before data set consisted of almost two years, or seven quarters
of prior data from September 2012 to September 2014, and one quarter of post
data (quarter four, 2014). The VSL project went into operation in September
2014. This limited the scope of the after data set to a single quarter, quarter
four, 2014. It was considered that three months of after data would be sufficient
for the effect of the implementation to stabilize. The following objectives were
defined for the evaluation:

• To demonstrate the power of data analytics on an application that is
within the state DOT’s current focus;

• To illustrate the application of external professional resources for data
analytics;
• To demonstrate how the application of analytics and predictive tech-
nologies can optimize the use of internal staff resources.

A review of previous evaluations of the VSL project suggests that there are
two primary approaches to the evaluation of VSLs. The first includes the use
of a traffic simulation model, and the second involves the analysis of before-
and-after data. Since the VSL project has already been deployed and one of the